Archive 4 checkpoints (1-4)

paper-releases/20260928/diffusion/data/cohorts/imperial_eskd/MANIFEST.json

@@ -0,0 +1,2090 @@
+{
+ "schema": "imperial_eskd_paired_ft_v1",
+ "written_utc": "2026-09-28T03:06:25.454996+00:00",
+ "started_utc": "2026-09-28T03:05:01.641260+00:00",
+ "study": "Imperial ESKD/haemodialysis COVID-19 cohort. Bulk: Gisby J et al., Nat Commun 2022 (Zenodo 10.5281/zenodo.6497251, PBMC RNA-seq, nf-core rnaseq 3.2 + htseq-count). scRNA: Stephenson E et al., Cell Genomics 2025 (COVID-19 Cell Atlas eskd_covid19.h5ad; PBMC CITE-seq GEX, 10x 5' v2, souporcell-demultiplexed pools)",
+ "rows": 172,
+ "row_order": "patient number ascending, then sample date",
+ "donors": 60,
+ "pairs_by_group": {
+  "covid_positive": 126,
+  "covid_negative": 36,
+  "recovery": 10
+ },
+ "donors_by_group": {
+  "covid_negative": 36,
+  "covid_positive": 38,
+  "recovery": 10
+ },
+ "rows_per_donor": {
+  "1": 26,
+  "2": 2,
+  "3": 16,
+  "4": 3,
+  "5": 4,
+  "6": 4,
+  "7": 2,
+  "8": 3
+ },
+ "pairs_by_basis": {
+  "patient+status+swab_day; symptom_day agrees": 110,
+  "patient+NEGATIVE; bulk metadata has no sample date (one NEGATIVE sample per patient on each side)": 36,
+  "patient+status+swab_day; neither side has a symptom-onset date": 26
+ },
+ "cells": {
+  "n_cells": 534014,
+  "nnz_panel6k": 196012727,
+  "per_row_min": 4,
+  "per_row_median": 2265.5,
+  "per_row_max": 11388,
+  "per_group_median": {
+   "covid_negative": 3015.5,
+   "covid_positive": 2136.5,
+   "recovery": 3135.5
+  }
+ },
+ "retained_cells_rule": "author-retained (the atlas already applies the authors' QC: exclude == 0 and souporcell singlet for every cell; >= 1000 UMIs, >= 200 genes, <= 10 % mito) AND flag_conservative_review_pass = genes >= 200 & UMIs >= 500 & mito <= 20 % (qc_core, ImMicro definition); MAD flags are review-only columns",
+ "features_semantics": {
+  "bulk_raw": "htseq-count raw gene counts on the canonical axis; 0 where bulk_valid is False (ImMicro file uses NaN there)",
+  "pseudo_raw_sum": "sum of retained cells' UMI counts on the canonical axis; storage 0 where sc_valid is False",
+  "bulk_valid": "harmonization_valid from qc_bulk.harmonize (gene in the htseq table, unique ID mapping)",
+  "sc_valid": "harmonization_valid from qc_singlecell.canonical_export (gene in the 10x GRCh38-3.0.0 feature table, unique ID mapping)",
+  "pair_valid": "bulk_valid & sc_valid",
+  "sample_ids": "sc author sample_id (patient_DDMM of the sample date); = sc_sample",
+  "donor_ids": "patient_id; several rows can share a donor (serial COVID-19 samples, pre-infection NEGATIVE, RECOVERY)",
+  "groups": "case_control: covid_positive (POSITIVE), covid_negative (NEGATIVE: pre-infection or uninfected control), recovery (RECOVERY)",
+  "timepoints": "days since the first positive swab (sc days_since_posswab = bulk time_from_first_positive_swab); NaN for covid_negative",
+  "severity": "WHO_severity_simple (peak severity over the illness: mild/moderate or severe/critical); 'none' for covid_negative (the atlas carries a placeholder there)",
+  "extra_keys": "bulk_sample (bulk column name), sc_sample, pairing_basis, sample_date, days_since_symptoms, bulk_day_label (the bulk column-name day = days since the earlier of symptom onset and positive swab), who_severity_peak, who_severity_at_sample, sex_meta, age_meta, centre, bulk_wave, identity_pass, identity_check, sex_bulk_expr, sex_sc_expr"
+ },
+ "cells_obs_semantics": {
+  "library_id": "sc author sample_id (cells from one sample span 2-4 pooled 10x libraries)",
+  "barcode": "atlas obs index (<pool GEX+ADT library IDs>_<10x barcode>), unique across the atlas",
+  "author_cell_type": "atlas cell_type_detailed",
+  "author_cell_type_broad": "atlas cell_type_broad (extra column)"
+ },
+ "identity": {
+  "sex_rule": "CPM on the source axis (bulk: all htseq genes; sc: retained cells, all RNA features); score=log2(XIST+1)-log2(RPS4Y1+DDX3Y+UTY+KDM5D+1); female>2, male<-2, else ambiguous (the Ye 2026 rule). sc only: where the pseudobulk call is ambiguous, the call falls back to per-cell detection (female if <20 % of retained cells have a Y-gene UMI and >=5 % have an XIST UMI; male if >=40 % have a Y-gene UMI and <20 % an XIST UMI), because the atlas X is pre-SoupX counts of pooled libraries whose ambient RNA carries the pool-mates' sex genes. Concordant = bulk call = sc call = metadata sex. Opposite-sex flag: female call with Y CPM>20 or male call with XIST CPM>50 (bulk)",
+  "expression_rule": "log2(CPM+1) over the 19656 canonical genes measured in all 336 bulks and all 198 sc samples; top 2000 by mean; Spearman of each paired bulk vs all sc pseudobulks and each paired sc vs all bulks; centred sensitivity",
+  "expression_residualised_rule": "within each modality: centre log2(CPM+1), remove the top k sample PCs (expressed genes: bulk mean >3 and sc mean >1 log2 CPM), keep the 2000 genes with the highest residual variance in both modalities (rank sum), Pearson; k=10 primary, 20/30 sensitivity. Uses no pairing information. Sample rank = rank of the exact partner; donor rank = rank of the own donor when every donor is scored by its best sample; visit rank = rank of the exact partner among the own donor's samples.",
+  "verdict_rule": "pass = sex concordant and residualised k=10 own DONOR ranked 1st in one direction and <=5th in the other; fail = sex discordant or own donor >5th in both directions; else borderline. The visit (time point) is not tested by the verdict: within a donor the residual profiles of different visits are close.",
+  "policy": "reported only; no row dropped on identity results; features.npz carries identity_pass / identity_check"
+ },
+ "checks": {
+  "gene_positions_sha256_ok": true,
+  "panel_gene_ids_match_selected_genes_csv": true,
+  "bulk_integer_nonnegative_inside_mask": true,
+  "pseudo_raw_sum_equals_sum_of_retained_cells_full_axis": true,
+  "cells_panel_sums_equal_pseudo_raw_sum_positions_exactly": true,
+  "cells_per_row_equal_qc_cell_counts": true,
+  "no_sc_mass_outside_sc_valid": true,
+  "one_sample_id_per_library": true,
+  "retained_cells_all_author_retained": true,
+  "sex_concordant_pairs": "169/172",
+  "expression_best_match_is_own_pair": "14/172",
+  "expression_best_match_is_own_pair_centred": "43/172",
+  "expression_reverse_best_match_is_own_pair": "4/172",
+  "expression_residualised_k10_bulk_to_sc_sample_top1": "130/172",
+  "expression_residualised_k10_sc_to_bulk_sample_top1": "140/172",
+  "expression_residualised_k10_bulk_to_sc_donor_top1": "158/172",
+  "expression_residualised_k10_sc_to_bulk_donor_top1": "159/172",
+  "expression_residualised_k20_bulk_to_sc_donor_top1": "159/172",
+  "expression_residualised_k30_bulk_to_sc_donor_top1": "157/172",
+  "expression_residualised_k10_exact_visit_top1_within_donor_bulk_to_sc": "115/150 rows whose donor has >1 sc sample",
+  "identity_check_pass": "158/172",
+  "identity_check_fail": [
+   "C33_2601",
+   "C137_1901",
+   "C139_3004",
+   "C139_2601",
+   "C139_2801",
+   "C139_0602",
+   "C139_0902",
+   "C224_2805"
+  ],
+  "identity_check_borderline": [
+   "C23_1004",
+   "C139_2301",
+   "C139_3001",
+   "C147_2301",
+   "C147_2801",
+   "C168_2601"
+  ],
+  "all_bulk_sex_call_differs_from_metadata": [
+   {
+    "bulk_sample": "C137_positive_2",
+    "donor_id": "C137",
+    "status": "positive",
+    "sex_meta": "F",
+    "bulk_sex_call": "ambiguous",
+    "bulk_XIST_cpm": 165.32725283513392,
+    "bulk_Ysum_cpm": 230.8627543149236,
+    "paired": false
+   },
+   {
+    "bulk_sample": "C137_positive_5",
+    "donor_id": "C137",
+    "status": "positive",
+    "sex_meta": "F",
+    "bulk_sex_call": "ambiguous",
+    "bulk_XIST_cpm": 157.77023622975042,
+    "bulk_Ysum_cpm": 175.55635451594068,
+    "paired": true
+   },
+   {
+    "bulk_sample": "C224_negative",
+    "donor_id": "C224",
+    "status": "negative",
+    "sex_meta": "M",
+    "bulk_sex_call": "ambiguous",
+    "bulk_XIST_cpm": 1092.210716744276,
+    "bulk_Ysum_cpm": 698.0330677203258,
+    "paired": true
+   },
+   {
+    "bulk_sample": "C33_positive_14",
+    "donor_id": "C33",
+    "status": "positive",
+    "sex_meta": "M",
+    "bulk_sex_call": "ambiguous",
+    "bulk_XIST_cpm": 74.86638377013038,
+    "bulk_Ysum_cpm": 197.04363178178681,
+    "paired": true
+   }
+  ],
+  "all_sc_sex_call_differs_from_metadata": []
+ },
+ "excluded": {
+  "quarantined_sc_samples_among_pairs": [],
+  "sc_samples_without_exact_bulk": [
+   {
+    "sc_sample": "C113_2304",
+    "patient_id": "C113",
+    "status": "positive",
+    "days_since_posswab": 4.0,
+    "sample_date": "23/04/2020",
+    "n_cells_atlas": 2165,
+    "reason": "no bulk on the same swab day (nearest 13 d)"
+   },
+   {
+    "sc_sample": "C113_2904",
+    "patient_id": "C113",
+    "status": "positive",
+    "days_since_posswab": 10.0,
+    "sample_date": "29/04/2020",
+    "n_cells_atlas": 5425,
+    "reason": "no bulk on the same swab day (nearest 7 d)"
+   },
+   {
+    "sc_sample": "C116_0205",
+    "patient_id": "C116",
+    "status": "positive",
+    "days_since_posswab": 12.0,
+    "sample_date": "02/05/2020",
+    "n_cells_atlas": 2648,
+    "reason": "no bulk on the same swab day (nearest 3 d)"
+   },
+   {
+    "sc_sample": "C116_2304",
+    "patient_id": "C116",
+    "status": "positive",
+    "days_since_posswab": 3.0,
+    "sample_date": "23/04/2020",
+    "n_cells_atlas": 1118,
+    "reason": "no bulk on the same swab day (nearest 12 d)"
+   },
+   {
+    "sc_sample": "C116_2804",
+    "patient_id": "C116",
+    "status": "positive",
+    "days_since_posswab": 8.0,
+    "sample_date": "28/04/2020",
+    "n_cells_atlas": 369,
+    "reason": "no bulk on the same swab day (nearest 7 d)"
+   },
+   {
+    "sc_sample": "C134_2904",
+    "patient_id": "C134",
+    "status": "negative",
+    "days_since_posswab": NaN,
+    "sample_date": "29/04/2020",
+    "n_cells_atlas": 1630,
+    "reason": "no bulk column for this patient and status"
+   },
+   {
+    "sc_sample": "C137_2801",
+    "patient_id": "C137",
+    "status": "positive",
+    "days_since_posswab": 14.0,
+    "sample_date": "28/01/2021",
+    "n_cells_atlas": 626,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   },
+   {
+    "sc_sample": "C138_2601",
+    "patient_id": "C138",
+    "status": "positive",
+    "days_since_posswab": 10.0,
+    "sample_date": "26/01/2021",
+    "n_cells_atlas": 3735,
+    "reason": "no bulk on the same swab day (nearest 3 d)"
+   },
+   {
+    "sc_sample": "C138_2801",
+    "patient_id": "C138",
+    "status": "positive",
+    "days_since_posswab": 12.0,
+    "sample_date": "28/01/2021",
+    "n_cells_atlas": 892,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   },
+   {
+    "sc_sample": "C141_0903",
+    "patient_id": "C141",
+    "status": "recovery",
+    "days_since_posswab": 52.0,
+    "sample_date": "09/03/2021",
+    "n_cells_atlas": 2573,
+    "reason": "no bulk column for this patient and status"
+   },
+   {
+    "sc_sample": "C141_2601",
+    "patient_id": "C141",
+    "status": "positive",
+    "days_since_posswab": 10.0,
+    "sample_date": "26/01/2021",
+    "n_cells_atlas": 959,
+    "reason": "no bulk on the same swab day (nearest 4 d)"
+   },
+   {
+    "sc_sample": "C147_2601",
+    "patient_id": "C147",
+    "status": "positive",
+    "days_since_posswab": 14.0,
+    "sample_date": "26/01/2021",
+    "n_cells_atlas": 2252,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   },
+   {
+    "sc_sample": "C169_2601",
+    "patient_id": "C169",
+    "status": "positive",
+    "days_since_posswab": 10.0,
+    "sample_date": "26/01/2021",
+    "n_cells_atlas": 1536,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   },
+   {
+    "sc_sample": "C169_3001",
+    "patient_id": "C169",
+    "status": "positive",
+    "days_since_posswab": 14.0,
+    "sample_date": "30/01/2021",
+    "n_cells_atlas": 4256,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   },
+   {
+    "sc_sample": "C170_2601",
+    "patient_id": "C170",
+    "status": "positive",
+    "days_since_posswab": 7.0,
+    "sample_date": "26/01/2021",
+    "n_cells_atlas": 1050,
+    "reason": "no bulk on the same swab day (nearest 3 d)"
+   },
+   {
+    "sc_sample": "C170_2801",
+    "patient_id": "C170",
+    "status": "positive",
+    "days_since_posswab": 9.0,
+    "sample_date": "28/01/2021",
+    "n_cells_atlas": 552,
+    "reason": "no bulk on the same swab day (nearest 5 d)"
+   },
+   {
+    "sc_sample": "C170_3001",
+    "patient_id": "C170",
+    "status": "positive",
+    "days_since_posswab": 11.0,
+    "sample_date": "30/01/2021",
+    "n_cells_atlas": 2466,
+    "reason": "no bulk on the same swab day (nearest 3 d)"
+   },
+   {
+    "sc_sample": "C187_2601",
+    "patient_id": "C187",
+    "status": "positive",
+    "days_since_posswab": 19.0,
+    "sample_date": "26/01/2021",
+    "n_cells_atlas": 377,
+    "reason": "no bulk on the same swab day (nearest 3 d)"
+   },
+   {
+    "sc_sample": "C196_2605",
+    "patient_id": "C196",
+    "status": "negative",
+    "days_since_posswab": NaN,
+    "sample_date": "26/05/2020",
+    "n_cells_atlas": 2398,
+    "reason": "no bulk column for this patient and status"
+   },
+   {
+    "sc_sample": "C20_1604",
+    "patient_id": "C20",
+    "status": "positive",
+    "days_since_posswab": 13.0,
+    "sample_date": "16/04/2020",
+    "n_cells_atlas": 5290,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   },
+   {
+    "sc_sample": "C23_1504",
+    "patient_id": "C23",
+    "status": "positive",
+    "days_since_posswab": 12.0,
+    "sample_date": "15/04/2020",
+    "n_cells_atlas": 873,
+    "reason": "no bulk on the same swab day (nearest 5 d)"
+   },
+   {
+    "sc_sample": "C58_2701",
+    "patient_id": "C58",
+    "status": "positive",
+    "days_since_posswab": 7.0,
+    "sample_date": "27/01/2021",
+    "n_cells_atlas": 852,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   },
+   {
+    "sc_sample": "C60_1604",
+    "patient_id": "C60",
+    "status": "positive",
+    "days_since_posswab": 2.0,
+    "sample_date": "16/04/2020",
+    "n_cells_atlas": 4955,
+    "reason": "no bulk on the same swab day (nearest 7 d)"
+   },
+   {
+    "sc_sample": "C60_2104",
+    "patient_id": "C60",
+    "status": "positive",
+    "days_since_posswab": 7.0,
+    "sample_date": "21/04/2020",
+    "n_cells_atlas": 3985,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   },
+   {
+    "sc_sample": "C73_2004",
+    "patient_id": "C73",
+    "status": "positive",
+    "days_since_posswab": 3.0,
+    "sample_date": "20/04/2020",
+    "n_cells_atlas": 336,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   },
+   {
+    "sc_sample": "C82_2004",
+    "patient_id": "C82",
+    "status": "positive",
+    "days_since_posswab": 6.0,
+    "sample_date": "20/04/2020",
+    "n_cells_atlas": 1057,
+    "reason": "no bulk on the same swab day (nearest 2 d)"
+   }
+  ],
+  "bulk_samples_without_sc": [
+   {
+    "bulk_sample": "C10_positive_18",
+    "pid": "C10",
+    "status": "positive",
+    "day_label": 18.0,
+    "time_from_first_positive_swab": 18.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C10_positive_21",
+    "pid": "C10",
+    "status": "positive",
+    "day_label": 21.0,
+    "time_from_first_positive_swab": 21.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C10_positive_30",
+    "pid": "C10",
+    "status": "positive",
+    "day_label": 30.0,
+    "time_from_first_positive_swab": 30.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C101_positive_6",
+    "pid": "C101",
+    "status": "positive",
+    "day_label": 6.0,
+    "time_from_first_positive_swab": 6.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C101_positive_14",
+    "pid": "C101",
+    "status": "positive",
+    "day_label": 14.0,
+    "time_from_first_positive_swab": 14.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C101_negative",
+    "pid": "C101",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1|w2"
+   },
+   {
+    "bulk_sample": "C105_negative",
+    "pid": "C105",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": -7.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C108_positive_7",
+    "pid": "C108",
+    "status": "positive",
+    "day_label": 7.0,
+    "time_from_first_positive_swab": 7.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C108_positive_12",
+    "pid": "C108",
+    "status": "positive",
+    "day_label": 12.0,
+    "time_from_first_positive_swab": 12.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C108_positive_16",
+    "pid": "C108",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 16.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C11_positive_13",
+    "pid": "C11",
+    "status": "positive",
+    "day_label": 13.0,
+    "time_from_first_positive_swab": 13.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C11_positive_17",
+    "pid": "C11",
+    "status": "positive",
+    "day_label": 17.0,
+    "time_from_first_positive_swab": 17.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C11_positive_19",
+    "pid": "C11",
+    "status": "positive",
+    "day_label": 19.0,
+    "time_from_first_positive_swab": 19.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C11_positive_21",
+    "pid": "C11",
+    "status": "positive",
+    "day_label": 21.0,
+    "time_from_first_positive_swab": 21.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C110_positive_11",
+    "pid": "C110",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C110_positive_14",
+    "pid": "C110",
+    "status": "positive",
+    "day_label": 14.0,
+    "time_from_first_positive_swab": 14.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C110_positive_2",
+    "pid": "C110",
+    "status": "positive",
+    "day_label": 2.0,
+    "time_from_first_positive_swab": 2.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C110_positive_4",
+    "pid": "C110",
+    "status": "positive",
+    "day_label": 4.0,
+    "time_from_first_positive_swab": 4.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C113_positive_27",
+    "pid": "C113",
+    "status": "positive",
+    "day_label": 27.0,
+    "time_from_first_positive_swab": 25.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C114_positive_13",
+    "pid": "C114",
+    "status": "positive",
+    "day_label": 13.0,
+    "time_from_first_positive_swab": 12.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C114_positive_16",
+    "pid": "C114",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 15.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C114_positive_24",
+    "pid": "C114",
+    "status": "positive",
+    "day_label": 24.0,
+    "time_from_first_positive_swab": 23.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C114_positive_27",
+    "pid": "C114",
+    "status": "positive",
+    "day_label": 27.0,
+    "time_from_first_positive_swab": 26.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C114_positive_31",
+    "pid": "C114",
+    "status": "positive",
+    "day_label": 31.0,
+    "time_from_first_positive_swab": 30.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C114_positive_5",
+    "pid": "C114",
+    "status": "positive",
+    "day_label": 5.0,
+    "time_from_first_positive_swab": 4.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C114_positive_9",
+    "pid": "C114",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 8.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C116_positive_16",
+    "pid": "C116",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 15.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C116_positive_26",
+    "pid": "C116",
+    "status": "positive",
+    "day_label": 26.0,
+    "time_from_first_positive_swab": 25.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C12_positive_11",
+    "pid": "C12",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 8.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C12_positive_17",
+    "pid": "C12",
+    "status": "positive",
+    "day_label": 17.0,
+    "time_from_first_positive_swab": 14.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C12_positive_19",
+    "pid": "C12",
+    "status": "positive",
+    "day_label": 19.0,
+    "time_from_first_positive_swab": 16.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C12_positive_22",
+    "pid": "C12",
+    "status": "positive",
+    "day_label": 22.0,
+    "time_from_first_positive_swab": 19.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C12_positive_24",
+    "pid": "C12",
+    "status": "positive",
+    "day_label": 24.0,
+    "time_from_first_positive_swab": 21.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C123_positive_25",
+    "pid": "C123",
+    "status": "positive",
+    "day_label": 25.0,
+    "time_from_first_positive_swab": 22.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C123_positive_35",
+    "pid": "C123",
+    "status": "positive",
+    "day_label": 35.0,
+    "time_from_first_positive_swab": 32.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C123_positive_10",
+    "pid": "C123",
+    "status": "positive",
+    "day_label": 10.0,
+    "time_from_first_positive_swab": 7.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C126_positive_9",
+    "pid": "C126",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 9.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C126_positive_11",
+    "pid": "C126",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C126_positive_5",
+    "pid": "C126",
+    "status": "positive",
+    "day_label": 5.0,
+    "time_from_first_positive_swab": 5.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C127_positive_11",
+    "pid": "C127",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 8.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C127_positive_16",
+    "pid": "C127",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 13.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C127_positive_23",
+    "pid": "C127",
+    "status": "positive",
+    "day_label": 23.0,
+    "time_from_first_positive_swab": 20.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C128_positive_16",
+    "pid": "C128",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 13.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C128_positive_23",
+    "pid": "C128",
+    "status": "positive",
+    "day_label": 23.0,
+    "time_from_first_positive_swab": 20.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C129_positive_13",
+    "pid": "C129",
+    "status": "positive",
+    "day_label": 13.0,
+    "time_from_first_positive_swab": 12.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C129_positive_3",
+    "pid": "C129",
+    "status": "positive",
+    "day_label": 3.0,
+    "time_from_first_positive_swab": 2.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C129_positive_6",
+    "pid": "C129",
+    "status": "positive",
+    "day_label": 6.0,
+    "time_from_first_positive_swab": 5.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C129_positive_8",
+    "pid": "C129",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 7.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C130_negative",
+    "pid": "C130",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C132_positive_12",
+    "pid": "C132",
+    "status": "positive",
+    "day_label": 12.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C132_positive_14",
+    "pid": "C132",
+    "status": "positive",
+    "day_label": 14.0,
+    "time_from_first_positive_swab": 13.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C132_positive_8",
+    "pid": "C132",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 7.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C105_positive_5",
+    "pid": "C105",
+    "status": "positive",
+    "day_label": 5.0,
+    "time_from_first_positive_swab": 2.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C105_positive_8",
+    "pid": "C105",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 5.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C105_positive_10",
+    "pid": "C105",
+    "status": "positive",
+    "day_label": 10.0,
+    "time_from_first_positive_swab": 7.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C105_positive_15",
+    "pid": "C105",
+    "status": "positive",
+    "day_label": 15.0,
+    "time_from_first_positive_swab": 12.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C105_positive_17",
+    "pid": "C105",
+    "status": "positive",
+    "day_label": 17.0,
+    "time_from_first_positive_swab": 14.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C105_positive_3",
+    "pid": "C105",
+    "status": "positive",
+    "day_label": 3.0,
+    "time_from_first_positive_swab": 0.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C137_positive_2",
+    "pid": "C137",
+    "status": "positive",
+    "day_label": 2.0,
+    "time_from_first_positive_swab": 2.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C137_positive_9",
+    "pid": "C137",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 9.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C137_positive_12",
+    "pid": "C137",
+    "status": "positive",
+    "day_label": 12.0,
+    "time_from_first_positive_swab": 12.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C137_positive_16",
+    "pid": "C137",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 16.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C138_recovery",
+    "pid": "C138",
+    "status": "recovery",
+    "day_label": NaN,
+    "time_from_first_positive_swab": 47.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C139_recovery",
+    "pid": "C139",
+    "status": "recovery",
+    "day_label": NaN,
+    "time_from_first_positive_swab": 54.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C141_negative",
+    "pid": "C141",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1|w2"
+   },
+   {
+    "bulk_sample": "C145_positive_11",
+    "pid": "C145",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C147_positive_8",
+    "pid": "C147",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 8.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C149_positive_14",
+    "pid": "C149",
+    "status": "positive",
+    "day_label": 14.0,
+    "time_from_first_positive_swab": 14.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C149_positive_3",
+    "pid": "C149",
+    "status": "positive",
+    "day_label": 3.0,
+    "time_from_first_positive_swab": 3.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C153_negative",
+    "pid": "C153",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C162_negative",
+    "pid": "C162",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C163_negative",
+    "pid": "C163",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C166_negative",
+    "pid": "C166",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C168_positive_12",
+    "pid": "C168",
+    "status": "positive",
+    "day_label": 12.0,
+    "time_from_first_positive_swab": 12.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C169_positive_12",
+    "pid": "C169",
+    "status": "positive",
+    "day_label": 12.0,
+    "time_from_first_positive_swab": 12.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C173_negative",
+    "pid": "C173",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C18_positive_14",
+    "pid": "C18",
+    "status": "positive",
+    "day_label": 14.0,
+    "time_from_first_positive_swab": 14.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C18_positive_15",
+    "pid": "C18",
+    "status": "positive",
+    "day_label": 15.0,
+    "time_from_first_positive_swab": 15.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C18_positive_17",
+    "pid": "C18",
+    "status": "positive",
+    "day_label": 17.0,
+    "time_from_first_positive_swab": 17.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C18_positive_19",
+    "pid": "C18",
+    "status": "positive",
+    "day_label": 19.0,
+    "time_from_first_positive_swab": 19.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C180_negative",
+    "pid": "C180",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C187_positive_9",
+    "pid": "C187",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 9.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C189_negative",
+    "pid": "C189",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C20_positive_16",
+    "pid": "C20",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 15.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C201_negative",
+    "pid": "C201",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C21_positive_26",
+    "pid": "C21",
+    "status": "positive",
+    "day_label": 26.0,
+    "time_from_first_positive_swab": 25.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C21_positive_10",
+    "pid": "C21",
+    "status": "positive",
+    "day_label": 10.0,
+    "time_from_first_positive_swab": 9.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C21_positive_16",
+    "pid": "C21",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 15.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C22_positive_8",
+    "pid": "C22",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 8.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C22_positive_11",
+    "pid": "C22",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C22_positive_13",
+    "pid": "C22",
+    "status": "positive",
+    "day_label": 13.0,
+    "time_from_first_positive_swab": 13.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C24_positive_8",
+    "pid": "C24",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 8.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C24_positive_11",
+    "pid": "C24",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C24_positive_13",
+    "pid": "C24",
+    "status": "positive",
+    "day_label": 13.0,
+    "time_from_first_positive_swab": 13.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C241_negative",
+    "pid": "C241",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C242_negative",
+    "pid": "C242",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C245_negative",
+    "pid": "C245",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C25_positive_14",
+    "pid": "C25",
+    "status": "positive",
+    "day_label": 14.0,
+    "time_from_first_positive_swab": 14.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C25_positive_16",
+    "pid": "C25",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 16.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C25_positive_18",
+    "pid": "C25",
+    "status": "positive",
+    "day_label": 18.0,
+    "time_from_first_positive_swab": 18.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C257_negative",
+    "pid": "C257",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C27_positive_6",
+    "pid": "C27",
+    "status": "positive",
+    "day_label": 6.0,
+    "time_from_first_positive_swab": 5.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C27_positive_8",
+    "pid": "C27",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 7.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C31_positive_8",
+    "pid": "C31",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 4.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C31_positive_13",
+    "pid": "C31",
+    "status": "positive",
+    "day_label": 13.0,
+    "time_from_first_positive_swab": 9.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C31_positive_15",
+    "pid": "C31",
+    "status": "positive",
+    "day_label": 15.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C32_positive_16",
+    "pid": "C32",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 16.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C33_negative",
+    "pid": "C33",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1|w2"
+   },
+   {
+    "bulk_sample": "C33_positive_11",
+    "pid": "C33",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C33_positive_16",
+    "pid": "C33",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 16.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C34_positive_9",
+    "pid": "C34",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 9.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C37_positive_9",
+    "pid": "C37",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 2.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C37_positive_12",
+    "pid": "C37",
+    "status": "positive",
+    "day_label": 12.0,
+    "time_from_first_positive_swab": 5.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C40_positive_21",
+    "pid": "C40",
+    "status": "positive",
+    "day_label": 21.0,
+    "time_from_first_positive_swab": 21.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C40_positive_31",
+    "pid": "C40",
+    "status": "positive",
+    "day_label": 31.0,
+    "time_from_first_positive_swab": 31.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C40_positive_11",
+    "pid": "C40",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C40_positive_14",
+    "pid": "C40",
+    "status": "positive",
+    "day_label": 14.0,
+    "time_from_first_positive_swab": 14.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C42_positive_4",
+    "pid": "C42",
+    "status": "positive",
+    "day_label": 4.0,
+    "time_from_first_positive_swab": 2.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C42_positive_7",
+    "pid": "C42",
+    "status": "positive",
+    "day_label": 7.0,
+    "time_from_first_positive_swab": 5.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C47_positive_9",
+    "pid": "C47",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 7.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C47_positive_13",
+    "pid": "C47",
+    "status": "positive",
+    "day_label": 13.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C47_positive_18",
+    "pid": "C47",
+    "status": "positive",
+    "day_label": 18.0,
+    "time_from_first_positive_swab": 16.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C48_positive_2",
+    "pid": "C48",
+    "status": "positive",
+    "day_label": 2.0,
+    "time_from_first_positive_swab": 0.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C48_positive_4",
+    "pid": "C48",
+    "status": "positive",
+    "day_label": 4.0,
+    "time_from_first_positive_swab": 2.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C50_positive_14",
+    "pid": "C50",
+    "status": "positive",
+    "day_label": 14.0,
+    "time_from_first_positive_swab": 13.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C50_positive_16",
+    "pid": "C50",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 15.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C50_positive_19",
+    "pid": "C50",
+    "status": "positive",
+    "day_label": 19.0,
+    "time_from_first_positive_swab": 18.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C50_positive_23",
+    "pid": "C50",
+    "status": "positive",
+    "day_label": 23.0,
+    "time_from_first_positive_swab": 22.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C58_positive_2",
+    "pid": "C58",
+    "status": "positive",
+    "day_label": 2.0,
+    "time_from_first_positive_swab": 2.0,
+    "waves": "w2"
+   },
+   {
+    "bulk_sample": "C59_positive_12",
+    "pid": "C59",
+    "status": "positive",
+    "day_label": 12.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C59_positive_16",
+    "pid": "C59",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 15.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C60_positive_29",
+    "pid": "C60",
+    "status": "positive",
+    "day_label": 29.0,
+    "time_from_first_positive_swab": 27.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C60_positive_33",
+    "pid": "C60",
+    "status": "positive",
+    "day_label": 33.0,
+    "time_from_first_positive_swab": 31.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C60_positive_37",
+    "pid": "C60",
+    "status": "positive",
+    "day_label": 37.0,
+    "time_from_first_positive_swab": 35.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C60_positive_11",
+    "pid": "C60",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 9.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C63_positive_8",
+    "pid": "C63",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 5.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C63_positive_15",
+    "pid": "C63",
+    "status": "positive",
+    "day_label": 15.0,
+    "time_from_first_positive_swab": 12.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C64_negative",
+    "pid": "C64",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C65_positive_23",
+    "pid": "C65",
+    "status": "positive",
+    "day_label": 23.0,
+    "time_from_first_positive_swab": 18.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C65_positive_27",
+    "pid": "C65",
+    "status": "positive",
+    "day_label": 27.0,
+    "time_from_first_positive_swab": 22.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C65_positive_36",
+    "pid": "C65",
+    "status": "positive",
+    "day_label": 36.0,
+    "time_from_first_positive_swab": 31.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C65_positive_15",
+    "pid": "C65",
+    "status": "positive",
+    "day_label": 15.0,
+    "time_from_first_positive_swab": 10.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C65_positive_20",
+    "pid": "C65",
+    "status": "positive",
+    "day_label": 20.0,
+    "time_from_first_positive_swab": 15.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C66_positive_5",
+    "pid": "C66",
+    "status": "positive",
+    "day_label": 5.0,
+    "time_from_first_positive_swab": 0.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C67_positive_3",
+    "pid": "C67",
+    "status": "positive",
+    "day_label": 3.0,
+    "time_from_first_positive_swab": 2.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C68_positive_5",
+    "pid": "C68",
+    "status": "positive",
+    "day_label": 5.0,
+    "time_from_first_positive_swab": 5.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C7_positive_9",
+    "pid": "C7",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 9.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C7_positive_11",
+    "pid": "C7",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C73_positive_9",
+    "pid": "C73",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 5.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C73_positive_14",
+    "pid": "C73",
+    "status": "positive",
+    "day_label": 14.0,
+    "time_from_first_positive_swab": 10.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C77_positive_11",
+    "pid": "C77",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 7.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C80_positive_16",
+    "pid": "C80",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 14.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C82_positive_8",
+    "pid": "C82",
+    "status": "positive",
+    "day_label": 8.0,
+    "time_from_first_positive_swab": 8.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C84_positive_3",
+    "pid": "C84",
+    "status": "positive",
+    "day_label": 3.0,
+    "time_from_first_positive_swab": 3.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C87_positive_13",
+    "pid": "C87",
+    "status": "positive",
+    "day_label": 13.0,
+    "time_from_first_positive_swab": 13.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C87_positive_9",
+    "pid": "C87",
+    "status": "positive",
+    "day_label": 9.0,
+    "time_from_first_positive_swab": 9.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C87_positive_11",
+    "pid": "C87",
+    "status": "positive",
+    "day_label": 11.0,
+    "time_from_first_positive_swab": 11.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C89_negative",
+    "pid": "C89",
+    "status": "negative",
+    "day_label": NaN,
+    "time_from_first_positive_swab": NaN,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C9_positive_12",
+    "pid": "C9",
+    "status": "positive",
+    "day_label": 12.0,
+    "time_from_first_positive_swab": 7.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C9_positive_17",
+    "pid": "C9",
+    "status": "positive",
+    "day_label": 17.0,
+    "time_from_first_positive_swab": 12.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C93_positive_10",
+    "pid": "C93",
+    "status": "positive",
+    "day_label": 10.0,
+    "time_from_first_positive_swab": 10.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C93_positive_15",
+    "pid": "C93",
+    "status": "positive",
+    "day_label": 15.0,
+    "time_from_first_positive_swab": 15.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C99_positive_13",
+    "pid": "C99",
+    "status": "positive",
+    "day_label": 13.0,
+    "time_from_first_positive_swab": 3.0,
+    "waves": "w1"
+   },
+   {
+    "bulk_sample": "C99_positive_16",
+    "pid": "C99",
+    "status": "positive",
+    "day_label": 16.0,
+    "time_from_first_positive_swab": 6.0,
+    "waves": "w1"
+   }
+  ]
+ },
+ "inputs": {
+  "gene_positions": {
+   "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/artifacts/panel_ladder_floor01_1m_20260923/tiers/6k/gene_positions.npy",
+   "sha256": "653ffa45ac519000baa41086c52408009391d6e68cfb218bc6a50d34163fcb24"
+  },
+  "selected_genes": {
+   "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/artifacts/panel_ladder_floor01_1m_20260923/tiers/6k/selected_genes.csv",
+   "sha256": "09248a3930e3fdefe1dbcd5d60cae4be81028526dcf097dd78169f5a415edfb6"
+  },
+  "canonical_genes": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/strict_qc_v1_20260913/reference/canonical_genes.tsv",
+   "sha256": "484893b6f6ab673566074e0d2f32c4ec6628ced63f7e709b38ec602cb3e776ab"
+  },
+  "bulk_harmonized": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/bulk/Zenodo_6497251_htseq_counts/harmonized.h5ad",
+   "sha256": "7d6d7a18d0f3bdbf88e8a4124c30fc94238cb537c48f8a3d8931bcb66458a152"
+  },
+  "sources": [
+   {
+    "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/datasets/Zenodo_6497251/raw/htseq_counts.csv",
+    "bytes": 54969689,
+    "sha256": "1351f4fa9abf97fa520313add67072f8ffdfb93278f527a404c334236fc1704a",
+    "acquisition_receipt": true,
+    "url": "https://zenodo.org/records/6497251/files/htseq_counts.csv?download=1",
+    "receipt_sha256": "1351f4fa9abf97fa520313add67072f8ffdfb93278f527a404c334236fc1704a",
+    "receipt_bytes": 54969689,
+    "receipt_completed_at": "2026-09-27T04:47:38.908172+00:00",
+    "matches_receipt": true
+   },
+   {
+    "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/datasets/Zenodo_6497251/metadata/w1_metadata.csv",
+    "bytes": 36172,
+    "sha256": "309e10792b6d6a77e995310f4e2ef64a14258801d9efff57db5bcbdc4461a851",
+    "acquisition_receipt": true,
+    "url": "https://zenodo.org/records/6497251/files/w1_metadata.csv?download=1",
+    "receipt_sha256": "309e10792b6d6a77e995310f4e2ef64a14258801d9efff57db5bcbdc4461a851",
+    "receipt_bytes": 36172,
+    "receipt_completed_at": "2026-09-27T04:45:42.243266+00:00",
+    "matches_receipt": true
+   },
+   {
+    "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/datasets/Zenodo_6497251/metadata/w2_metadata.csv",
+    "bytes": 19690,
+    "sha256": "88a875406595bfebc5e163bb54261e6f528aaac23db0700170f20bcad042304b",
+    "acquisition_receipt": true,
+    "url": "https://zenodo.org/records/6497251/files/w2_metadata.csv?download=1",
+    "receipt_sha256": "88a875406595bfebc5e163bb54261e6f528aaac23db0700170f20bcad042304b",
+    "receipt_bytes": 19690,
+    "receipt_completed_at": "2026-09-27T04:45:40.345447+00:00",
+    "matches_receipt": true
+   },
+   {
+    "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/datasets/ESKD_COVID19_CellAtlas/raw/eskd_covid19.h5ad",
+    "bytes": 3814688826,
+    "sha256": "e3cc87788bdc212b9cbb3c1f4c56e89146751fdcdf3366a9b67f84f898b93408",
+    "acquisition_receipt": true,
+    "url": "https://covid19.cog.sanger.ac.uk/eskd_covid19.h5ad",
+    "receipt_sha256": "e3cc87788bdc212b9cbb3c1f4c56e89146751fdcdf3366a9b67f84f898b93408",
+    "receipt_bytes": 3814688826,
+    "receipt_completed_at": "2026-09-27T05:29:49.522136+00:00",
+    "matches_receipt": true
+   }
+  ],
+  "gene_id_crosswalk": [
+   {
+    "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/reference_crosswalk/cellranger_GRCh38-3.0.0_features__from_GSE161529.tsv.gz",
+    "sha256": "d8b21cc094b58a3bdcd6c2b3cc7699399c4f14ff8f8611231a2c6da04791a354",
+    "bytes": 270380
+   },
+   {
+    "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/reference_crosswalk/cellranger_GRCh38-3.0.0_features__from_GSM8012326.tsv.gz",
+    "sha256": "41314847b784a19147284b0f15d4e4ce9ee5ad4b0a277c300af1e477c4629ba4",
+    "bytes": 304728
+   }
+  ]
+ },
+ "code": {
+  "assemble": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/imperial_assemble.py",
+   "sha256": "9192cb45806dca35924cab86ad764ccc1db4387a95ac41818545794c0b4c5ba0"
+  },
+  "pipeline": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/imperial_pipeline.py",
+   "sha256": "c5b6f56a2b6770f8d3d7d1a0298c1f75b2ee72a4d14ed35233aa7d875eee6535"
+  },
+  "run_step": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/run_step.sh",
+   "sha256": "5b79a56423056a4229ee26bd37a10179a460a02b64e1be1046d33a1f51837df6"
+  },
+  "bundle/qc_bulk.py": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/strict_qc_v1_code_bundle_v5/qc_bulk.py",
+   "sha256": "8c3bd5179a11c25fdeea491d7071bd492faaabc3cff3348a0750de199cddae9c"
+  },
+  "bundle/qc_core.py": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/strict_qc_v1_code_bundle_v5/qc_core.py",
+   "sha256": "b50a175b64044142676c8c55f9630da90d8c0c193c761124de9ef02d517e4851"
+  },
+  "bundle/qc_finalize.py": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/strict_qc_v1_code_bundle_v5/qc_finalize.py",
+   "sha256": "c4acc068779e3823869bafa4cb8a9384b60b3e66693bbacc313758d0bd495f0a"
+  },
+  "bundle/qc_genes.py": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/strict_qc_v1_code_bundle_v5/qc_genes.py",
+   "sha256": "ff93bbc364af6a0822b0cff7c5ef0fb407aa2512bc8afa6f0681754f0fb65c86"
+  },
+  "bundle/qc_singlecell.py": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/strict_qc_v1_code_bundle_v5/qc_singlecell.py",
+   "sha256": "a69ab1e0b1686e466474735b67a398f5d2140936e66a95051d02a6a497a59753"
+  },
+  "bundle/qc_validate_outputs.py": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/strict_qc_v1_code_bundle_v5/qc_validate_outputs.py",
+   "sha256": "8fe39a62b38f40e3e6053e9300abe81818a27f0e39a044e660eb88c094480daa"
+  },
+  "bundle/strict_qc_driver.py": {
+   "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/strict_qc_v1_code_bundle_v5/strict_qc_driver.py",
+   "sha256": "88ec34fb10074334a4e4ba2f4059a9b04ade1ba3b335d2c448a1f67594daa078"
+  }
+ },
+ "processing_dir": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928",
+ "findings": {
+  "bulk_column_day": {
+   "positive_columns": 269,
+   "label_equals_days_since_positive_swab": 132,
+   "label_equals_days_since_first_symptoms": 189,
+   "label_equals_max_of_both": 269,
+   "conclusion": "The bulk column-name day is days since the earlier of symptom onset and positive swab (max of the two metadata clocks) for all 269 COVID-19-positive columns; it is NOT days since positive swab. Pairing therefore uses the metadata swab clock (time_from_first_positive_swab == sc days_since_posswab), checked against the symptom clock.",
+   "coordinator_quick_rule": {
+    "rule": "bulk column-name day == sc days_since_posswab (POSITIVE only)",
+    "exact_sc_samples": 76,
+    "exact_patients": 24,
+    "nearest_offset_1_day": 30,
+    "nearest_offset_2_days": 15,
+    "why_it_differs": "the column-name day counts from the earlier of symptom onset and positive swab, days_since_posswab counts from the swab; they differ whenever symptoms preceded the swab"
+   }
+  },
+  "gene_ids": {
+   "atlas_var": "Seurat make.unique symbols only (33,538 human + 21 VIRAL- features)",
+   "resolution": "Ensembl IDs restored by position from the 10x Cell Ranger GRCh38-3.0.0 features.tsv (exact ordered match of all 33,538 names); see processing reference_crosswalk/CROSSWALK.json",
+   "canonical_genes_measured_sc": 19656,
+   "canonical_genes_measured_bulk": 19954,
+   "panel6k_measured_sc": 5927,
+   "panel6k_measured_bulk": 5980,
+   "panel6k_measured_both": 5927,
+   "symbol_only_mapping_would_differ_at_features": 494
+  },
+  "atlas_vs_paper": {
+   "atlas": {
+    "cells": 588389,
+    "samples": 198,
+    "patients": 63
+   },
+   "paper": {
+    "cells": 580040,
+    "samples": 187,
+    "patients": 61
+   },
+   "author_filters_already_applied": {
+    "exclude_nonzero_cells": 0,
+    "non_singlet_cells": 0,
+    "min_nCount_RNA_raw": 1005.0,
+    "min_nFeature_RNA_raw": 200.0,
+    "max_percentMito": 9.998711727976367
+   },
+   "patients_without_any_b_cell": [
+    "C116",
+    "C139"
+   ],
+   "their_samples": [
+    "C116_0205",
+    "C116_2304",
+    "C116_2804",
+    "C139_0602",
+    "C139_0902",
+    "C139_2301",
+    "C139_2601",
+    "C139_2801",
+    "C139_3001",
+    "C139_3004"
+   ],
+   "bulk_ROR1_cpm_median_by_patient": {
+    "C116": 18.241555526240226,
+    "C139": 46.36882994994504
+   },
+   "bulk_ROR1_cpm_cohort_median": 0.2804417972378572,
+   "samples_below_10_cells": [
+    "C168_2601"
+   ],
+   "interpretation": "The paper removed 'samples from two individuals ... with benign chronic lymphocytic leukemia'. The atlas keeps 63 patients, but two of them have no B cell at all and their bulk is ROR1-high (a CLL marker): most likely the atlas dropped their B cells, not the patients. 198 - their 10 samples - 1 sample with 4 cells = 187, the paper's sample count (cells do not reconcile exactly). Inference, not author-confirmed."
+  },
+  "sc_sex_ambient": {
+   "rule": "see identity.sex_rule",
+   "samples_called_per_cell": [
+    "C145_3004",
+    "C169_0202",
+    "C169_2301",
+    "C23_1004",
+    "C23_1504",
+    "C85_2004"
+   ],
+   "male_meta_frac_cells_Y_min": 0.5,
+   "female_meta_frac_cells_Y_max": 0.07698289269051321
+  },
+  "bulk_opposite_sex_signal": [
+   {
+    "bulk_sample": "C137_positive_2",
+    "bulk_XIST_cpm": 165.3,
+    "bulk_Ysum_cpm": 230.9,
+    "sex_meta": "F",
+    "paired": false
+   },
+   {
+    "bulk_sample": "C137_positive_5",
+    "bulk_XIST_cpm": 157.8,
+    "bulk_Ysum_cpm": 175.6,
+    "sex_meta": "F",
+    "paired": true
+   },
+   {
+    "bulk_sample": "C224_negative",
+    "bulk_XIST_cpm": 1092.2,
+    "bulk_Ysum_cpm": 698.0,
+    "sex_meta": "M",
+    "paired": true
+   },
+   {
+    "bulk_sample": "C33_positive_14",
+    "bulk_XIST_cpm": 74.9,
+    "bulk_Ysum_cpm": 197.0,
+    "sex_meta": "M",
+    "paired": true
+   }
+  ],
+  "sc_chemistry": {
+   "paper": "Chromium Next GEM Single Cell 5' Kit v2 (CITE-seq TotalSeq-C; TCR/BCR); Cell Ranger 4.0, GRCh38-3.0.0 reference",
+   "first_sorted_barcode_prefix_per_pool": {
+    "AAACCTG": 116,
+    "AAATGCC": 1,
+    "AAAGCAA": 1
+   },
+   "note": "AAACCTG first barcodes = 737K-august-2016 whitelist, consistent with 5' chemistry"
+  },
+  "mito_percent_vs_author": {
+   "max_abs_diff_per_sample_median": 0.3552408776275513,
+   "max": 3.998130494545462,
+   "note": "qc_core mito % (13 MT genes / all RNA UMIs, raw counts) vs author percentMito; no cell exceeds 20 % either way"
+  },
+  "downstream_compatibility": [
+   "donor_ids repeat (serial samples): prepare_dataset folds by donor and handles this; tools/ft_common.load_features refuses repeated donor_ids ('this scorer scores one sample per donor').",
+   "groups covid_negative / covid_positive / recovery: ft_common.CONTROL_RE does not match 'covid_negative', so an automatic case-vs-control contrast finds no control group."
+  ]
+ },
+ "outputs": {
+  "cells_obs.parquet": {
+   "bytes": 15396124,
+   "sha256": "58d685b6b9510cd07bcabe61f90a1567b9d483efe3c1c035305fd3525682aa97"
+  },
+  "cells_panel6k.npz": {
+   "bytes": 363733249,
+   "sha256": "d648fa1b9efa15f02c5500c85795d33c39cc88857560bf8f0aae41f6d43a3de7"
+  },
+  "features.npz": {
+   "bytes": 12991654,
+   "sha256": "e46e5bb1a24eae045efe75125d20e3accaa8f68a1006eb2f60d24159013d4aa7"
+  },
+  "identity_expression.csv": {
+   "bytes": 73070,
+   "sha256": "9bc9216a0d30a999a3da61d77d42214cd05353c971512275e39f6f8b66a3f373"
+  },
+  "identity_sex_bulk_all.csv": {
+   "bytes": 57766,
+   "sha256": "0ead468fedc94b85fbc8f56ee98cd5f9eab3d9d9c20540feb4e27a8341e61805"
+  },
+  "identity_sex_sc_all.csv": {
+   "bytes": 44399,
+   "sha256": "4fd4a029bc0122bc3b551affcf0c76e96b935f0be13cf9b10ea319705db9acb5"
+  },
+  "pairs.csv": {
+   "bytes": 147906,
+   "sha256": "cc97eedb9dd86114c75ba553b2d8c0deb0879945bcd0aaca3d6c064ff23e9d3d"
+  },
+  "pairs_sensitivity_offday.csv": {
+   "bytes": 1483,
+   "sha256": "54a1b5937784f37a59e2c9e4ea3e76b4b0f81fe6a396c02f6fd8745e4245f3c9"
+  },
+  "samples.csv": {
+   "bytes": 6448,
+   "sha256": "6b2fdfd57e2b60065f399f8727311279a2a1d91cab28945c5e9445c55a5e69d6"
+  }
+ },
+ "post_assembly": {
+  "written_utc": "2026-09-28T03:08:51.442439+00:00",
+  "cells_full": {
+   "dataset": "imperial_eskd",
+   "n_cells": 534014,
+   "nnz": 683805349,
+   "sha256": "4629d8cc14d1a6b73f7c5e28847ffce215ad316055b2bde8e90f7ca2a4bd6f67",
+   "checks": {
+    "panel_columns_equal_cells_panel6k": true,
+    "row_sums_equal_pseudo_raw_sum": true
+   },
+   "builder": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/imperial_full_cells.py",
+   "builder_sha256": "b3aa12fadf51c2e34809a34bd87b7d0fc9df5b6b2604ac55073f60aae37e0af2",
+   "bytes": 5476852390
+  },
+  "prepare_dataset": {
+   "command": "CUDA_VISIBLE_DEVICES='' python tools/prepare_dataset.py --dataset imperial_eskd",
+   "tool_sha256": "ec12a021808bcc120b382bd4b7cdefb60a470ff760c1aa0cba6425aee01bb28e",
+   "PREPARE.json_sha256": "0957cd872133ae8c87f0364cf5b18c0b216099778c4b4b7af2642754a431b838",
+   "folds.csv_sha256": "18c6619ab176ec01040be372761e6280da75acebc6be9b9408b526b3e07a7e55",
+   "fold_sizes": {
+    "0": {
+     "test": 44,
+     "train": 100,
+     "inner_val": 14,
+     "test_groups": {
+      "covid_positive": 35,
+      "covid_negative": 7,
+      "recovery": 2
+     },
+     "alpha_genes_estimated": 10957
+    },
+    "1": {
+     "test": 25,
+     "train": 112,
+     "inner_val": 21,
+     "test_groups": {
+      "covid_positive": 18,
+      "covid_negative": 7
+     },
+     "alpha_genes_estimated": 10954
+    },
+    "2": {
+     "test": 30,
+     "train": 107,
+     "inner_val": 21,
+     "test_groups": {
+      "covid_positive": 21,
+      "covid_negative": 7,
+      "recovery": 2
+     },
+     "alpha_genes_estimated": 10971
+    },
+    "3": {
+     "test": 29,
+     "train": 116,
+     "inner_val": 13,
+     "test_groups": {
+      "covid_positive": 19,
+      "covid_negative": 7,
+      "recovery": 3
+     },
+     "alpha_genes_estimated": 10970
+    },
+    "4": {
+     "test": 30,
+     "train": 108,
+     "inner_val": 20,
+     "test_groups": {
+      "covid_positive": 21,
+      "covid_negative": 6,
+      "recovery": 3
+     },
+     "alpha_genes_estimated": 10979
+    }
+   },
+   "rows_in_fold_minus1": 14,
+   "donors_in_folds": 58,
+   "donors_split_across_folds": 0
+  },
+  "code": {
+   "full_cells": {
+    "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/imperial_full_cells.py",
+    "sha256": "b3aa12fadf51c2e34809a34bd87b7d0fc9df5b6b2604ac55073f60aae37e0af2"
+   },
+   "finalize": {
+    "path": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/imperial_finalize.py",
+    "sha256": "9995db3f7acaa3b29d57abfef19c2201bf9b3bcaa7e0327e2ed0093d78a920d4"
+   }
+  }
+ }
+}

paper-releases/20260928/diffusion/data/cohorts/imperial_eskd/PREPARE.json

@@ -0,0 +1,95 @@
+{
+ "schema": "pbisc_sdd.ft_intradataset.conditions.v1",
+ "dataset": "imperial_eskd",
+ "written_utc": "2026-09-28T03:12:44.848197+00:00",
+ "features_sha256": "e46e5bb1a24eae045efe75125d20e3accaa8f68a1006eb2f60d24159013d4aa7",
+ "seed": 20260927,
+ "folds": 5,
+ "inner_frac": 0.1,
+ "cells_per_bulk": 1500,
+ "recipe_k": 1000,
+ "depth_anchor": 3869.603120208939,
+ "builder": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/realbulk_generation_20260922/build_realbulk_conditions.py",
+ "builder_sha256": "e818d69e6801b5699579db7fc450590a266708d1fc2d544743e8624c9ee6a30a",
+ "files": {
+  "conditions_raw_fixed.npz": "41a62291e98e367d05108a0a95a859ef9ca1997f1fda4656cba58eeefbe56442",
+  "conditions_pseudo_fixed.npz": "94998a1852dc1dd867179f9b34ebdd36ed7405ce084ca63aad41f2222e397b3e",
+  "fold0/conditions_cal_fixed.npz": "c50bfab9ace76854dbd6c9effcd815aac9739a37fdb2e97b2ab50cff6f4cc4c2",
+  "fold0/test_raw.npz": "131ad25e4bff3dfc3e542b9a69bd0c3a4ab9979a14f159788d6df9b024271805",
+  "fold0/test_cal.npz": "5afee50728707cba9a617b90adfb3283d9819c828e1bcaa3ddece75c16b59d45",
+  "fold0/test_pseudo.npz": "ae27831a54e4f670ebf3199d842edafc59bf50a38a355545ea7b2149a4e27e17",
+  "fold1/conditions_cal_fixed.npz": "3e6afdc0bd3bd15ee604872c176e428b427bfbf9904cfde361a06ca325b9195d",
+  "fold1/test_raw.npz": "e5e19d453e39ba25206139a4321e1812d5c6bca9018e68c8a565476b51a275e5",
+  "fold1/test_cal.npz": "3c4c1ad14e485fe49fc05cf69e4809bbb4e7355c3b8b67440e2fd9f16632e151",
+  "fold1/test_pseudo.npz": "e1e863d9a34d2313ef77ff6ac355edab5babbfd2c88cf847d2d8d493f708861d",
+  "fold2/conditions_cal_fixed.npz": "99d821522cf845e655fb584f99a4dfbdc84f50d102f11a710fab9ec9be70237d",
+  "fold2/test_raw.npz": "c394f98b927e9788417f5151af520441b9f2b2abfee331298deb9d049e0fa1a7",
+  "fold2/test_cal.npz": "669624734ba1443e7ad446fb39f36043330c2d13e6ece1c7e5b584f001ed687c",
+  "fold2/test_pseudo.npz": "4d191352498cabc825d1919005e2e9ecf048eb8cc42087bc22bc13e3d69ec254",
+  "fold3/conditions_cal_fixed.npz": "d368c0e0f14c20102f24954b5c2ed80a0926b2d5ae0a75eddde002083ce489a4",
+  "fold3/test_raw.npz": "3fd0c680cd90fcf9bb6aeb52112a4c8c690a9b7483cb834a14dcf0e9f7700b6f",
+  "fold3/test_cal.npz": "ae3c023ca4043713782a308b5e4dfeb712cddf56c784a732c459528337a7a268",
+  "fold3/test_pseudo.npz": "90897bdf2384d6b63c624f4eb17bd6b022510c9c2c940e76291310752e41146d",
+  "fold4/conditions_cal_fixed.npz": "e00a24686f1a068d5a91ae0196bedbbf869d4db28b4e91095c106e49d5f80470",
+  "fold4/test_raw.npz": "19b375355dc21cc38127ab631e74c1346909707524598abe164bc887f7018253",
+  "fold4/test_cal.npz": "896acfbf9095e014c0d30c20c1668fe9a5c35c30cca288bcf8548cfb63203a4a",
+  "fold4/test_pseudo.npz": "3abd907d08b17c0704f84c30f01002231ef24417771ebd98a475c52844d3588d"
+ },
+ "fold_sizes": {
+  "0": {
+   "test": 47,
+   "train": 97,
+   "inner_val": 13,
+   "test_groups": {
+    "covid_positive": 37,
+    "covid_negative": 7,
+    "recovery": 3
+   },
+   "alpha_genes_estimated": 10932
+  },
+  "1": {
+   "test": 25,
+   "train": 121,
+   "inner_val": 11,
+   "test_groups": {
+    "covid_positive": 18,
+    "covid_negative": 7
+   },
+   "alpha_genes_estimated": 10967
+  },
+  "2": {
+   "test": 29,
+   "train": 121,
+   "inner_val": 7,
+   "test_groups": {
+    "covid_positive": 20,
+    "covid_negative": 7,
+    "recovery": 2
+   },
+   "alpha_genes_estimated": 10960
+  },
+  "3": {
+   "test": 24,
+   "train": 115,
+   "inner_val": 18,
+   "test_groups": {
+    "covid_positive": 15,
+    "covid_negative": 7,
+    "recovery": 2
+   },
+   "alpha_genes_estimated": 10973
+  },
+  "4": {
+   "test": 32,
+   "train": 111,
+   "inner_val": 14,
+   "test_groups": {
+    "covid_positive": 23,
+    "covid_negative": 6,
+    "recovery": 3
+   },
+   "alpha_genes_estimated": 10980
+  }
+ },
+ "checks": {}
+}

paper-releases/20260928/diffusion/data/cohorts/imperial_eskd/QC_REPORT.md

@@ -0,0 +1,140 @@
+# Imperial ESKD COVID-19: paired PBMC bulk and scRNA (QC report)
+
+Study: haemodialysis (end-stage kidney disease) patients with COVID-19, Imperial cohort.
+- Bulk: Gisby et al., Nat Commun 2022. Zenodo 6497251, PBMC RNA-seq (nf-core rnaseq 3.2, then htseq-count).
+- scRNA: Stephenson et al., Cell Genomics 2025. COVID-19 Cell Atlas `eskd_covid19.h5ad`, PBMC CITE-seq, gene expression only (ADT not used).
+
+Built 2026-09-28 on server_1, CPU only. Processing directory: `/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/`.
+
+## Result
+
+**172 pairs from 60 patients.** One row is one sc sample with its bulk sample from the same patient and day. **158 pass** the identity checks, **8 fail** and **6 are borderline**. The rows are kept and flagged (`identity_pass`); `prepare_dataset.py` puts the 14 non-passing rows in fold −1.
+
+| group | pairs | patients | pass | retained cells per sample, median (min to max) |
+|---|---|---|---|---|
+| covid_positive | 126 | 38 | 114 | 2,137 (4 to 9,254) |
+| covid_negative | 36 | 36 | 34 | 3,016 (265 to 11,388) |
+| recovery | 10 | 10 | 10 | 3,136 (587 to 5,693) |
+| all | 172 | 60 | 158 | 2,266 (4 to 11,388); 534,014 cells in total |
+
+Patients have 1 to 8 rows (26 patients have 1 row). Donor = patient, so `donor_ids` repeat.
+
+## Pairing rule
+
+**The day in the bulk column name is not days since positive swab.**
+- In all 269 COVID-19-positive bulk columns, the name's day equals max(`time_from_first_symptoms`, `time_from_first_positive_swab`). That is days since the earlier of symptom onset and the positive swab.
+- It equals the swab day in only 132 columns.
+- The quick rule "column day == sc `days_since_posswab`" therefore finds 76 exact pairs from 24 patients. It also puts 30 at 1 day off and 15 at 2 days off. Those off-by-one and off-by-two cases are the same days counted from different clocks.
+
+**Rule used:**
+- **POSITIVE and RECOVERY:** same patient and status, and bulk `time_from_first_positive_swab` == sc `days_since_posswab`. Both count from the same patient's swab date, so the two samples come from the same calendar day.
+  - Where both sides have a symptom-onset day, it must agree too. It agrees in all 110 such pairs, with no contradictions.
+  - 26 pairs have no symptom date on either side and rest on the swab day alone.
+- **NEGATIVE:** same patient, one NEGATIVE sample on each side (36 pairs). The bulk metadata has no date for NEGATIVE samples, so these pairs are patient + status, not proven same draw.
+- Every pair is 1:1. Patient sex and age agree between the bulk metadata and the atlas for all 63 patients.
+
+**Sensitivity list** (`pairs_sensitivity_offday.csv`): 13 candidates for 10 unpaired sc samples, all at ±2 days on the swab clock. There are no ±1-day candidates. Five of the 13 bulks already belong to a primary pair. These candidates are not in the primary files.
+
+**Not paired:**
+- 26 sc samples. For 23 the nearest same-patient bulk is 2 to 13 days away. The other 3 have no bulk for that patient and status: 2 NEGATIVE, whose bulk IDs appear in the metadata but not in the count matrix, and 1 RECOVERY.
+- 164 bulk samples have no sc.
+
+Lists are in MANIFEST.json (`excluded`) and in `processing/.../pairing/`.
+
+## Identity checks
+
+**Sex.** XIST against RPS4Y1 + DDX3Y + UTY + KDM5D in CPM on the source axis, using the Ye 2026 rule.
+- 169 of 172 pairs agree across bulk, sc and metadata.
+- **One change for sc.** The atlas X is pre-SoupX counts of pooled 10x libraries. Ambient RNA from male pool-mates raises the Y signal in female samples (up to 66 CPM), so five female samples came out "ambiguous" at pseudobulk level.
+  - For those five, the call uses per-cell detection instead.
+  - In every metadata-female sample at most 7.7 % of cells have any Y UMI. In every metadata-male sample at least 50 % do.
+  - The five samples: C23_1004, C85_2004, C145_3004, C169_2301, C169_0202.
+- **Three bulks show opposite-sex signal**, and their pairs fail:
+  - C33_positive_14: male, XIST 75 CPM.
+  - C137_positive_5: female, Y 176 CPM. C137's other draws are clean except positive_2.
+  - C224_negative: male, XIST 1,092 CPM and Y 698 CPM.
+  - Bulk libraries are not pooled, so this points to mixed material.
+- Twelve passing pairs have a small opposite-sex bulk signal (`bulk_possible_mixture`).
+
+**Expression, donor-sensitive.** This decides pass or fail and uses no pairing information.
+- Method: within each modality, remove the top 10 sample PCs (all 336 bulks, all 198 sc samples). Then take the 2,000 genes with the highest remaining variance in both modalities and use Pearson correlation.
+- The own patient is the top match in 158 of 172 rows (bulk to sc) and 159 of 172 (sc to bulk). The result is the same with 20 or 30 PCs removed (159 and 157).
+- The exact sample is the top match in 130 and 140 rows. Within a patient, the exact visit ranks first in 115 of 150 rows whose patient has more than one sc sample. Visits of one patient look alike after the PC removal, so the verdict tests the donor, not the day.
+- Passing rows have r = 0.47 (median).
+
+**Verdict rule.** Pass means:
+- sex agrees; and
+- the own patient ranks 1st in one direction and within the top 5 in the other. Each patient is represented by its best-matching sample.
+
+| row | bulk | verdict | reason |
+|---|---|---|---|
+| C139_3004, C139_2601, C139_2801, C139_0602, C139_0902 | C139 | fail | own patient ranks 10 to 39. See C139 below. |
+| C139_2301, C139_3001 | C139 | borderline | own patient ranks 15–16 in one direction; 61 and 195 cells |
+| C33_2601 | C33_positive_14 | fail | bulk sex ambiguous |
+| C137_1901 | C137_positive_5 | fail | bulk sex ambiguous |
+| C224_2805 | C224_negative | fail | bulk sex ambiguous (both XIST and Y high) |
+| C147_2301 | C147_positive_11 | borderline | own patient ranks 19 and 1 |
+| C147_2801 | C147_positive_16 | borderline | own patient ranks 6 and 5 |
+| C23_1004 | C23_positive_8 | borderline | own patient ranks 2 and 3 |
+| C168_2601 | C168_positive_10 | borderline | only 4 cells; ranks 1 and 43 |
+
+**C139 (and C116, which has no pair).** These are the only two patients with no B cells at all in the atlas. Their bulk is ROR1-high: C139 26 to 90 CPM and C116 18 CPM, against a cohort median of 0.3. C139's bulk CD79A is 10 times the median.
+
+The paper removed "samples from two individuals ... with benign chronic lymphocytic leukemia". The likely explanation is that the atlas dropped these patients' B cells but kept the patients. Two numbers support this:
+- 198 samples, minus their 10, minus one 4-cell sample, gives exactly the paper's 187.
+- The atlas has 63 patients and the paper 61.
+
+For C139 the bulk contains B cells that the sc does not, so the pair cannot match. This is an inference, not author-confirmed.
+
+## Processing
+
+The run used the frozen strict-QC code (bundle v5, copied unchanged and sha-checked), the same 20,097-gene GENCODE v49 axis and the same mapping rules.
+
+**Bulk**
+- 60,649 versioned GENCODE gene IDs × 336 samples. All values are nonnegative integers, and there are no htseq `__` rows.
+- 19,954 canonical genes are present and 143 are missing (marked unmeasured).
+- 44 `_PAR_Y` rows are quarantined and 107 IDs are unknown.
+- On the 6k panel, 5,980 of 6,000 genes are measured.
+
+**scRNA**
+- One pooled H5AD: 588,389 cells, 198 samples, 63 patients, 118 10x pools demultiplexed by souporcell. Each sample spans 2 or 4 pools.
+- Each sample was cut out of its pools and run through the bundle as one library, as for Ye and ImMicro.
+- **Counts are raw UMIs before SoupX.** For every cell, the X row sum equals the author column `nCount_RNA_raw` and the number of stored entries equals `nFeature_RNA_raw`. The SoupX-corrected `nCount_RNA` is not what X holds. The `logcounts` layer was not used.
+- **Gene IDs.** The atlas var holds only Seurat `make.unique` symbols.
+  - Their order matches the 10x Cell Ranger GRCh38-3.0.0 `features.tsv` at all 33,538 positions, including the 24 renamed duplicates. That is the reference the paper states.
+  - Ensembl IDs were taken from that table by position; the proof is in `processing/.../reference_crosswalk/CROSSWALK.json`. Mapping is then by ID, as in ImMicro.
+  - 19,656 canonical genes are present and 441 are missing. On the 6k panel, **5,927 of 6,000 genes are measured**; 5,927 are measured in both assays.
+  - 19 genes have a symbol that disagrees with the ID and are kept under the ID.
+  - A symbol-only mapping would differ at 494 features.
+- The 21 viral features (77 UMIs in total) are outside the axis.
+- **Cells.** The atlas already applies the authors' QC to every cell: `exclude` is 0, the souporcell call is singlet, UMIs are at least 1,005, genes at least 200 and mito at most 10 %. So all 588,389 cells also pass `flag_conservative_review_pass` (at least 200 genes, at least 500 UMIs, at most 20 % mito). The 5-MAD flags are review-only columns.
+- Retained cells in the 172 paired samples: 534,014.
+- 34 rows have fewer than 1,000 cells and 6 have fewer than 200 (`below_200_cells`). One of the six, C73_2404 with 179 cells, passes identity.
+- Chemistry: 10x 5′ v2 (paper). 116 of 118 pools start with the AAACCTG barcode prefix, which is consistent with 5′ chemistry.
+- `author_cell_type` is `cell_type_detailed`. `author_cell_type_broad` is an extra column.
+
+## Files
+
+| file | contents |
+|---|---|
+| `features.npz` | One row per pair, ordered by patient then sample date. Core keys follow Ye 2026. Extra keys: `timepoints`, `severity`, `bulk_sample`, `sc_sample`, `pairing_basis`, `below_200_cells`. `bulk_raw` is 0 where not measured. |
+| `cells_panel6k.npz` | CSR int32 counts for the retained cells: 534,014 × 6,000, 196.0 M nonzero. |
+| `cells_full.npz` | The same cells on the full 20,097 axis: CSR int32, 683.8 M nonzero, 5.5 GB. |
+| `cells_obs.parquet` | One row per retained cell, same order. |
+| `samples.csv`, `pairs.csv` | Row index to sample and group; pairing fields, sex and expression checks, and the verdict. |
+| `pairs_sensitivity_offday.csv` | ±2-day candidates, not used. |
+| `identity_*.csv` | Full sex tables (all 336 bulks and 198 sc samples) and expression ranks. |
+| `MANIFEST.json` | Sources with sha256, code with sha256, checks, findings, cells_full and the prepare_dataset run. |
+
+Checks passed:
+- For every row, the summed panel counts equal `pseudo_raw_sum[:, positions]` exactly. On the full axis, the summed counts equal `pseudo_raw_sum`.
+- Cell counts equal `qc_cell_counts`, and there are no counts outside the measured masks.
+- The sha256 of `gene_positions` matches.
+- `tools/prepare_dataset.py --dataset imperial_eskd` ran. It made 5 donor-level folds over 58 patients. No patient is split across folds, and the 14 non-passing rows are in fold −1.
+
+## Notes for downstream use
+
+- `tools/ft_common.load_features` refuses repeated `donor_ids` ("this scorer scores one sample per donor"). Scoring this dataset needs a per-sample mode.
+- `covid_negative` does not match `ft_common.CONTROL_RE`, so the automatic case-vs-control contrast finds no control group.
+- Fold sizes differ in rows (25 to 44 test rows) because folds balance patients, and patients have 1 to 8 rows.
+- NEGATIVE pairs are the same patient and status but not proven same draw.

paper-releases/20260928/diffusion/data/cohorts/imperial_eskd/cells_full.BUILD.json

@@ -0,0 +1,12 @@
+{
+ "dataset": "imperial_eskd",
+ "n_cells": 534014,
+ "nnz": 683805349,
+ "sha256": "4629d8cc14d1a6b73f7c5e28847ffce215ad316055b2bde8e90f7ca2a4bd6f67",
+ "checks": {
+  "panel_columns_equal_cells_panel6k": true,
+  "row_sums_equal_pseudo_raw_sum": true
+ },
+ "builder": "/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/imperial_eskd_v1_20260928/code/imperial_full_cells.py",
+ "builder_sha256": "b3aa12fadf51c2e34809a34bd87b7d0fc9df5b6b2604ac55073f60aae37e0af2"
+}