Dataset card revisions

README.md

@@ -82,54 +82,40 @@ for `leakage_free`, and 13,449 pairs over 298 targets for `leaked`. It contains
 `predict_seconds`.
 
 **`esmc_metrics`** contains ESMC [1] scores for the same pairs as
-`esmfold2_metrics`, over two feature corpora, so 48,044 rows in total. Its score
-column is `npmi_score`, alongside `set`, `feature_corpus`, `method`, `target`,
+`esmfold2_metrics`, with the ESMC features normalized over two different feature corpora (see below). 48,044 rows in total. Its score column is `npmi_score`, alongside `set`, `feature_corpus`, `method`, `target`,
 `partner`, `pair_key`, `n_residues`, `type`, `status` and `imputed_zero`.
 `status` is `OK` on every row.
 
-ESMC scores a pair from the normalised pointwise mutual information (nPMI)
+ESMC scores a pair from the normalised pointwise mutual information (NPMI)
 between the sparse autoencoder features of the two proteins, counted over a
-corpus of PDB structures [1]. That corpus is itself a leakage surface, so the
-scores are published over two forms of it under `feature_corpus`:
+corpus of PDB structures [1]. We provide ESMC scores normalized against two different feature corpora, the PPI3D dataset with two different date cutoffs labelled in `feature_corpus`:
 
-- `full` — the whole corpus, 631,509 feature pairs.
+- `full` — the corpus up to 2026-07-29, 631,509 feature pairs.
 - `pre_2023-06-01` — the corpus restricted to structures released before the
-  benchmark cutoff, 510,811 feature pairs. This is ESMC's own leakage control,
-  and the counterpart of the `leakage_free` / `leaked` split for a method whose
-  leakage runs through co-occurrence statistics rather than through weights.
+  benchmark cutoff, 510,811 feature pairs. This allows fair comparisons to ESMFold2 on the Leakage-Free set
 
-Each corpus covers the same 24,022 pairs, so `set` × `feature_corpus` is a
-complete 2×2 and the four cells can be read directly against one another.
-
-426 pairs per corpus (188 `leakage_free` and 238 `leaked`) are scored `0` and
+Each corpus covers the exact same 24,022 pairs. 426 pairs per corpus (188 `leakage_free` and 238 `leaked`) are scored `0` and
 marked `imputed_zero`: ESMC found no residue in one of the two proteins whose
-features passed the activation threshold. Nothing failed there — extraction ran
-to completion and the model's answer is that it found nothing — so those pairs
-sit at the bottom of the ranking rather than outside the file. The same pairs are
-imputed under both corpora, because a protein with no active features has none in
-either. Filter on `imputed_zero` to drop them instead.
+features passed the activation threshold.
 
 **`zhang_pairs`**, **`zhang_sequences`** and **`zhang_esmc_metrics`** are an
-external test set and our result on it, rather than a slice of Flock: they share
-no pairs, targets or leakage flags with any file above.
+external test set and our result on it, they share no pairs, targets or leakage flags with any file above.
 
 `zhang_pairs` is the human PPI control set of Zhang et al. [15], parsed from
 their Data S1. Their positives are 3,000 pairs sampled from the 3,988 confident
 PPIs that STRING, BioGRID and UniProt all agree on; their negatives are 30,000
 random human protein pairs with no evidence of interaction. We then filtered that
-set for homology against the PPI3D [6] structure manifest behind the nPMI feature
-table, with MMseqs2 [16] at 30% sequence identity and 80% coverage. 6,409 of its
+set for homology against the PPI3D [6] structures that fed into our NPMI feature
+table, using MMseqs2 clustering [16] at 30% sequence identity and 80% coverage. 6,409 of its
 17,266 proteins matched the corpus at that threshold, taking it from 32,747 pairs
 (3,000 positive and 29,747 negative) to the 12,635 published here: 738 positive
 and 11,897 negative over 9,820 proteins. It carries `protein_a`, `protein_b` and
 `type`, with the accessions sorted as in `flock_pairs`.
 
 `zhang_sequences` gives `accession`, `sequence` and `length` for all 9,820
-proteins. These are the exact sequences that were scored, taken from the run's
-own input rather than refetched from UniProt, so they cannot drift from the
-scores published beside them.
+proteins, the exact sequences that were scored.
 
-`zhang_esmc_metrics` holds ESMC's nPMI score over that set against the `full`
+`zhang_esmc_metrics` holds ESMC's NPMI score over that set against the `full`
 feature corpus, one row per pair. Alongside `npmi_score` it carries the per-side
 active feature counts `n_active_a`, `n_active_b`, `n_selected` and `n_positive`,
 plus `status` and `imputed_zero`. Pooled AUROC is 0.9015 imputing the 67
@@ -144,6 +130,8 @@ no-feature pairs to zero, or 0.9010 dropping them.
 > where the AUROC above is unweighted over a 738:11,897 mix. Read it as ESMC as we
 > ran it, on a homology-filtered subset of their control set.
 
+> **Reproduction note** We endeavoured to reproduce the published ESMC method as best as possible; however some details and parameters were not described in their paper. In particular the exact date cutoff behind their feature corpus is not provided; our PPI3D cutoff of 2026-07-29 may be slightly later than theirs and as such our NPMI set may be slightly bigger. In our paper we report comparable performance on this set.
+
 ## Loading
 
 To train models on Flock, load the data using HuggingFace`Datasets`:
@@ -282,7 +270,7 @@ The sources carry their own terms, all of which permit redistribution here:
 | UniProt [11] | CC BY 4.0 |
 | AlphaFold DB [12, 13] | CC BY 4.0 |
 
-The pairs in `zhang_pairs` and `zhang_esmc_metrics` are a filtered subset of the control sets in Data S1 of Zhang et al. [15], redistributed here with attribution so that the scores published beside them can be checked. Those control sets are themselves drawn from STRING, BioGRID and UniProt and from random sampling of the human proteome, and what appears here is 12,635 of their 32,747 pairs: accessions and labels only, none of the predicted structures or interaction probabilities carried by the paper's other supplementary files. The accepted manuscript is open in PMC (PMC13281779); the published article is subscription.
+The pairs in `zhang_pairs` and `zhang_esmc_metrics` are a filtered subset of the control sets in Data S1 of Zhang et al. [15], redistributed here with attribution so that the scores published beside them can be checked.
 
 Two additional sources were used to generate this dataset but do not appear in it. **IntAct** [8] is used only as a filter, to remove pairs with curated evidence of interaction from the negatives, so no IntAct records are redistributed here. **Negatome 2.0** [14] inspired this paper, and while this work builds on the methods used there, the data presented here is generated from different inputs, sources, and code.