junidude14 commited on
Commit
2abf739
·
verified ·
1 Parent(s): 737a48f

Archive 4 checkpoints (1-4)

Browse files
paper-releases/pbisc-diffusion/updates/20261013T012029Z-restore-original-deck-style/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/data/style_review_20261013/REPORT.md ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Restore the original 76-slide figure style
2
+
3
+ User directive: Figure 1 and the review deck were visually inferior; match the existing 76-slide deck.
4
+
5
+ ## Cause and correction
6
+
7
+ The prior builder introduced DejaVu Sans, teal headings, tinted status cards, smaller plot fonts and crops of the manuscript overview draft. This departed from the supplied deck. The revised builder calls the original slide renderer directly: Arial, 26-pt title, grey top-right labels, white background, original margins, 16-pt takeaways and 9-pt footnotes. The original 76-slide deck and all its source assets remain unchanged.
8
+
9
+ Figure 1 restores original slides 7, 8 and 9: the complete biological overview, bulk-conditioning diagram and corpus composition. All inserted figure images are uncropped. Original diagram semantics are preserved; the caption distinguishes conditioning recipe size from generated-cell count.
10
+
11
+ Six panels were redrawn only from saved tables using the original shared style: ablation, full-FT/LoRA rank, healthy composition, observed rare states, corrected rare-state comparison and Behcet C1Q rules. Method colors match the original: PBISC LoRA #D55E00 and scSemiProfiler #E69F00. No new training, generation or scoring was executed.
12
+
13
+ ## Evidence wording retained or corrected
14
+
15
+ - Existing exact-five healthy results retain their original protocol.
16
+ - Historical learning curves retain n training + 7 validation. scSemi budget curves also disclose their CV-HVG reuse; they are not presented as strict total-n experiments.
17
+ - Cohort identity screening is described as available by cohort, because ImMicro had no identity flag.
18
+ - Disease five-donor sample budgets, rare-state diagnostic status and unmatched Behcet limitations remain visible.
19
+ - Figure 6 remains waiting for user-prepared data.
20
+
21
+ ## Validation
22
+
23
+ Original renderer/style/deck hashes are recorded in originals_before.json. VERIFICATION.json checks all 34 slides, PDF pages, text/shape bounds, Arial slide typography, source hashes and byte-identical uncropped reused images. All-page contact sheets and full-size selected slide renders were visually inspected. The two plotting modules received independent visual reviews. Healthy composition and observed rare-state matrices exactly match the prior saved values, including missingness. Rare comparison retains 59 abundance pairs and 53 finite ranking pairs.
24
+
25
+ The archive policy receives the exact changed reusable assets and this record; no original 76-slide file or frozen release is rewritten.
paper-releases/pbisc-diffusion/updates/20261013T012029Z-restore-original-deck-style/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/data/style_review_20261013/VERIFICATION.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema": "pbisc.deck-style-verification.v1",
3
+ "slides": 34,
4
+ "pdf_pages": 34,
5
+ "style_renderer": "original generate_deck.py",
6
+ "fonts": [
7
+ "Arial"
8
+ ],
9
+ "reused_original_figure_images_verified": 26,
10
+ "all_figures_uncropped": true,
11
+ "source_hashes_verified": 99,
12
+ "originals_preserved": 6,
13
+ "shape_and_pdf_text_bounds_pass": true,
14
+ "pdf_sha256": "9e22d2e9bb2f37b6f8d43ef758c276860365a526dbdce08c1cdf255095884c97",
15
+ "pptx_sha256": "a17d5ebfd73dcfea2f70f85d7f3529ffa8b4330b03968c59e9ee737f687f1690",
16
+ "new_training": false,
17
+ "new_generation": false,
18
+ "numerical_changes": false
19
+ }
paper-releases/pbisc-diffusion/updates/20261013T012029Z-restore-original-deck-style/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.provenance.json ADDED
@@ -0,0 +1,236 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema": "pbisc.saved-table-style.v1",
3
+ "panel": "behcet",
4
+ "operation": "visual restyling of saved estimates; no new experiment or scoring",
5
+ "style": {
6
+ "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/tools/style.py",
7
+ "sha256": "0d17caf276ae0e9b25285b8fd8f9b75c55621ad3a36382b705a366f327c89de6",
8
+ "bytes": 3468
9
+ },
10
+ "renderer": {
11
+ "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/tools/results_review_style_panels.py",
12
+ "sha256": "395662295dd314369711123ecadda7ec5faa7cc916b55cdf2ba3f55307cc56cd",
13
+ "bytes": 15521
14
+ },
15
+ "sources": [
16
+ {
17
+ "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/rare_behcet_per_sample.csv",
18
+ "sha256": "d19b5007de6861ba30730c4e15c90c78bf4d5e1f115ea9464ac66adbe5e2722e",
19
+ "bytes": 18038
20
+ },
21
+ {
22
+ "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/reports/rare_behcet_provenance.json",
23
+ "sha256": "f2147fb96c914cd957853b151191fec004079066dd7ab017e16b508512f6554f",
24
+ "bytes": 1204
25
+ }
26
+ ],
27
+ "outputs": {
28
+ "png": {
29
+ "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.png",
30
+ "sha256": "692ecf5ac216c661f465cb9b534e5809978fdfcd8dfc82376b7d1517093ee9b0",
31
+ "bytes": 190359
32
+ },
33
+ "pdf": {
34
+ "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.pdf",
35
+ "sha256": "1ec32fd693caf48e5a66fc48944fd473d6bb424a03574e93168722b0d32427b7",
36
+ "bytes": 33260
37
+ },
38
+ "svg": {
39
+ "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.svg",
40
+ "sha256": "a26515b1b9f1267b22478851d5e685378ae5c99bc139e294a2a5455cb4eb3b57",
41
+ "bytes": 49470
42
+ }
43
+ },
44
+ "scope": "Descriptive unmatched Behcet C1Q A/B comparison",
45
+ "group_means": [
46
+ {
47
+ "arm": "obs",
48
+ "group": "HC",
49
+ "rule": "A",
50
+ "n_samples": 4,
51
+ "mean_percent": 6.833102241576405
52
+ },
53
+ {
54
+ "arm": "obs",
55
+ "group": "BD",
56
+ "rule": "A",
57
+ "n_samples": 4,
58
+ "mean_percent": 8.157227537984546
59
+ },
60
+ {
61
+ "arm": "obs_thin_lora3",
62
+ "group": "HC",
63
+ "rule": "A",
64
+ "n_samples": 4,
65
+ "mean_percent": 6.70367844054749
66
+ },
67
+ {
68
+ "arm": "obs_thin_lora3",
69
+ "group": "BD",
70
+ "rule": "A",
71
+ "n_samples": 4,
72
+ "mean_percent": 7.964010680166572
73
+ },
74
+ {
75
+ "arm": "zs_raw_s13",
76
+ "group": "HC",
77
+ "rule": "A",
78
+ "n_samples": 10,
79
+ "mean_percent": 3.0981382715176453
80
+ },
81
+ {
82
+ "arm": "zs_raw_s13",
83
+ "group": "BD",
84
+ "rule": "A",
85
+ "n_samples": 6,
86
+ "mean_percent": 2.1583367449262183
87
+ },
88
+ {
89
+ "arm": "zs_cal_s13",
90
+ "group": "HC",
91
+ "rule": "A",
92
+ "n_samples": 10,
93
+ "mean_percent": 3.4256958893505725
94
+ },
95
+ {
96
+ "arm": "zs_cal_s13",
97
+ "group": "BD",
98
+ "rule": "A",
99
+ "n_samples": 6,
100
+ "mean_percent": 4.540172174688849
101
+ },
102
+ {
103
+ "arm": "lora3_raw_s13",
104
+ "group": "HC",
105
+ "rule": "A",
106
+ "n_samples": 10,
107
+ "mean_percent": 3.948593545731355
108
+ },
109
+ {
110
+ "arm": "lora3_raw_s13",
111
+ "group": "BD",
112
+ "rule": "A",
113
+ "n_samples": 6,
114
+ "mean_percent": 4.859228750378392
115
+ },
116
+ {
117
+ "arm": "lora3_raw_s04",
118
+ "group": "HC",
119
+ "rule": "A",
120
+ "n_samples": 10,
121
+ "mean_percent": 3.55366247827943
122
+ },
123
+ {
124
+ "arm": "lora3_raw_s04",
125
+ "group": "BD",
126
+ "rule": "A",
127
+ "n_samples": 6,
128
+ "mean_percent": 5.2750592287064935
129
+ },
130
+ {
131
+ "arm": "obs",
132
+ "group": "HC",
133
+ "rule": "B",
134
+ "n_samples": 4,
135
+ "mean_percent": 2.4456667729132326
136
+ },
137
+ {
138
+ "arm": "obs",
139
+ "group": "BD",
140
+ "rule": "B",
141
+ "n_samples": 4,
142
+ "mean_percent": 3.0295354560220775
143
+ },
144
+ {
145
+ "arm": "obs_thin_lora3",
146
+ "group": "HC",
147
+ "rule": "B",
148
+ "n_samples": 4,
149
+ "mean_percent": 2.4073819337095577
150
+ },
151
+ {
152
+ "arm": "obs_thin_lora3",
153
+ "group": "BD",
154
+ "rule": "B",
155
+ "n_samples": 4,
156
+ "mean_percent": 2.8832359093405424
157
+ },
158
+ {
159
+ "arm": "zs_raw_s13",
160
+ "group": "HC",
161
+ "rule": "B",
162
+ "n_samples": 10,
163
+ "mean_percent": 0.199728700857221
164
+ },
165
+ {
166
+ "arm": "zs_raw_s13",
167
+ "group": "BD",
168
+ "rule": "B",
169
+ "n_samples": 6,
170
+ "mean_percent": 0.11096827206164
171
+ },
172
+ {
173
+ "arm": "zs_cal_s13",
174
+ "group": "HC",
175
+ "rule": "B",
176
+ "n_samples": 10,
177
+ "mean_percent": 0.307816748287003
178
+ },
179
+ {
180
+ "arm": "zs_cal_s13",
181
+ "group": "BD",
182
+ "rule": "B",
183
+ "n_samples": 6,
184
+ "mean_percent": 0.34788103912347995
185
+ },
186
+ {
187
+ "arm": "lora3_raw_s13",
188
+ "group": "HC",
189
+ "rule": "B",
190
+ "n_samples": 10,
191
+ "mean_percent": 0.24144936880229398
192
+ },
193
+ {
194
+ "arm": "lora3_raw_s13",
195
+ "group": "BD",
196
+ "rule": "B",
197
+ "n_samples": 6,
198
+ "mean_percent": 0.25416353448912
199
+ },
200
+ {
201
+ "arm": "lora3_raw_s04",
202
+ "group": "HC",
203
+ "rule": "B",
204
+ "n_samples": 10,
205
+ "mean_percent": 0.352271096064559
206
+ },
207
+ {
208
+ "arm": "lora3_raw_s04",
209
+ "group": "BD",
210
+ "rule": "B",
211
+ "n_samples": 6,
212
+ "mean_percent": 0.3722319033847383
213
+ }
214
+ ],
215
+ "evidence": {
216
+ "schema": "pbisc.fig5.behcet-descriptive.v1",
217
+ "sources": {
218
+ "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/behcet_c1q_lora3_20261004/out/rules_groups/per_sample_group_metrics.csv": "1cee6722b93ca58bd077e79fe63d2b47aac1a4a2ac80a854f6722c5c386f6ff1",
219
+ "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/behcet_pair_recovery_20261004/out/pairs_recovered.csv": "25ec1ea12d625547b63d705200d79675cc6124c594e2f91c4dd588adc87345f1",
220
+ "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/behcet_c1q_lora3_20261004/out/group_table.csv": "55c00f70aeb523039203a8cf4cb8b55ff4e0fcbbed348db69179a833f46eb40b"
221
+ },
222
+ "training_bulk_samples": [
223
+ "BDV4",
224
+ "BDV5",
225
+ "BDV6"
226
+ ],
227
+ "pair_identity": "2 primary passes; 1 uncertain",
228
+ "calibration": "zs_cal is gene-wise calibration fitted on the same three candidate pairs; it is not external AIFI ref-cal",
229
+ "real_reference": "4 BD and 4 HC; includes training sc donors, mostly unmatched to held-out bulks",
230
+ "rules": {
231
+ "A": "at least one C1Q gene detected",
232
+ "B": "at least two C1Q genes co-detected in the same cell"
233
+ },
234
+ "independent_paired_validation": false
235
+ }
236
+ }
paper-releases/pbisc-diffusion/updates/20261013T012029Z-restore-original-deck-style/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.svg ADDED