Upload folder using huggingface_hub

.gitattributes

@@ -393,3 +393,18 @@ analysis/replogle_source_to_filtered_schema/replogle_source_filtered_schema_no_x
 analysis/replogle_source_to_filtered_schema/replogle_source_filtered_schema_with_xstate.h5ad filter=lfs diff=lfs merge=lfs -text
 analysis/replogle_xstate_reembed/replogle_concat_input_no_xstate.h5ad filter=lfs diff=lfs merge=lfs -text
 analysis/replogle_xstate_reembed/replogle_concat_reembedded_se600m.h5ad filter=lfs diff=lfs merge=lfs -text
+data/adamson_state_compatible/processed/adamson_state_schema_no_xstate.h5ad filter=lfs diff=lfs merge=lfs -text
+data/adamson_state_compatible/processed/adamson_state_schema_with_xstate.h5ad filter=lfs diff=lfs merge=lfs -text
+data/adamson_state_compatible/processed/state_tx_data/adamson.h5ad filter=lfs diff=lfs merge=lfs -text
+data/adamson_state_compatible/raw/Adamson.h5ad filter=lfs diff=lfs merge=lfs -text
+data/replogle_nadig_hf/replogle.h5ad filter=lfs diff=lfs merge=lfs -text
+data/replogle_nadig_hf/rpe1_normalized_singlecell_01.h5ad filter=lfs diff=lfs merge=lfs -text
+data/replogle_nogwps_v2/rpe1.h5ad filter=lfs diff=lfs merge=lfs -text
+data/replogle_nogwps_v2_rpe1_checkpoint_vocab/rpe1.h5ad filter=lfs diff=lfs merge=lfs -text
+data/replogle_self_se/rpe1_raw_from_replogle.h5ad filter=lfs diff=lfs merge=lfs -text
+data/replogle_self_se/rpe1_self_se_embedded.h5ad filter=lfs diff=lfs merge=lfs -text
+data/replogle_self_se_for_st/rpe1.h5ad filter=lfs diff=lfs merge=lfs -text
+data/replogle_source_filtered_schema_for_st/replogle_concat.h5ad filter=lfs diff=lfs merge=lfs -text
+data/state_replogle_filtered/replogle_concat.h5ad filter=lfs diff=lfs merge=lfs -text
+data/state_replogle_filtered_analysis/replogle_concat.h5ad filter=lfs diff=lfs merge=lfs -text
+data/state_replogle_filtered_reembedded/replogle_concat.h5ad filter=lfs diff=lfs merge=lfs -text

data/adamson_state_compatible/processed/adamson_state_schema_no_xstate.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6af0a00914837dab88599477a5b481ea49cbd95ea2c318326f8d5bcaf614069d
+size 610539813

data/adamson_state_compatible/processed/adamson_state_schema_with_xstate.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8005bf133336aaefb5bf55b958a2b6a7571e08dde60287bbe2299581c0139ae
+size 3201233772

data/adamson_state_compatible/processed/state_tx_data/adamson.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a05c8631414cdc78cd7cf88dc0302e783e34a2bd20f834a24b1256566bd74e4
+size 1035695957

data/adamson_state_compatible/raw/Adamson.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54c626f7f9dab83a68ae066cccba2d7fc7f8a6be679ccceff518234afb9dde09
+size 2275857650

data/adamson_state_compatible/raw/Adamson.h5ad.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5694f4972a8df13d1127624180daaaeb02a88488b0494737aa36df04c1cddcc
+size 369137596

data/adamson_state_compatible/raw/data/adamson_state_compatible/patch_adamson_to_state_schema.py

@@ -0,0 +1,219 @@
+import os
+from pathlib import Path
+
+import anndata as ad
+import numpy as np
+import pandas as pd
+import scipy.sparse as sp
+
+src = Path(os.environ["ADAMSON_RAW_H5AD"])
+dst = Path(os.environ["ADAMSON_NO_STATE_H5AD"])
+report_dir = Path(os.environ["ADAMSON_REPORT_DIR"])
+
+dst.parent.mkdir(parents=True, exist_ok=True)
+report_dir.mkdir(parents=True, exist_ok=True)
+
+print("Reading:", src)
+a = ad.read_h5ad(src)
+
+print("\nRAW ANNData")
+print(a)
+print("obs columns:", list(a.obs.columns))
+print("var columns:", list(a.var.columns))
+print("layers:", list(a.layers.keys()))
+print("obsm:", list(a.obsm.keys()))
+
+# ---------------------------------------------------------------------
+# 1. Confirm expression matrix.
+# ---------------------------------------------------------------------
+# Your inspection showed:
+#   X == layer["logNor"]
+#   layer["counts"] is raw counts
+# Therefore we keep a.X as the SE input matrix.
+if "logNor" in a.layers:
+    x_sample = a.X[:1000, :1000]
+    l_sample = a.layers["logNor"][:1000, :1000]
+
+    if sp.issparse(x_sample):
+        x_sample = x_sample.toarray()
+    if sp.issparse(l_sample):
+        l_sample = l_sample.toarray()
+
+    mad = float(np.mean(np.abs(np.asarray(x_sample) - np.asarray(l_sample))))
+    print("\nX vs layer['logNor'] sample mean abs diff:", mad)
+
+    if mad > 1e-8:
+        print("WARNING: X differs from layer['logNor']; keeping X as-is.")
+else:
+    print("\nNo layer['logNor'] found; keeping X as-is.")
+
+# ---------------------------------------------------------------------
+# 2. Ensure gene names are usable.
+# ---------------------------------------------------------------------
+# Adamson var_names appear to be the expression gene names.
+# Ensembl IDs are stored separately in var["Ensembl_ID"].
+a.var_names_make_unique()
+
+print("\nvar_names sample:")
+print(list(a.var_names[:20]))
+
+if "Ensembl_ID" in a.var.columns:
+    print("\nEnsembl_ID sample:")
+    print(a.var["Ensembl_ID"].astype(str).head(20).to_list())
+
+# ---------------------------------------------------------------------
+# 3. Perturbation column -> obs["gene"].
+# ---------------------------------------------------------------------
+# Adamson already has obs["gene"], with CTRL as the control.
+if "gene" not in a.obs.columns:
+    raise ValueError("Expected obs['gene'] in Adamson file, but it is missing.")
+
+a.obs["gene_original"] = a.obs["gene"].astype(str)
+a.obs["gene"] = a.obs["gene"].astype(str)
+
+print("\nTop raw perturbations:")
+print(a.obs["gene"].value_counts().head(30))
+
+# Normalize controls to the STATE-style label.
+control_aliases = {
+    "CTRL",
+    "ctrl",
+    "Control",
+    "control",
+    "CONTROL",
+    "NT",
+    "NTC",
+    "non-targeting",
+    "non_targeting",
+    "nontargeting",
+    "negative_control",
+    "negctrl",
+    "mock",
+    "unperturbed",
+}
+
+mask_control = a.obs["gene"].astype(str).str.lower().isin(
+    {x.lower() for x in control_aliases}
+)
+a.obs.loc[mask_control, "gene"] = "non-targeting"
+
+print("\nTop perturbations after control normalization:")
+print(a.obs["gene"].value_counts().head(30))
+print("non-targeting cells:", int((a.obs["gene"] == "non-targeting").sum()))
+
+if int((a.obs["gene"] == "non-targeting").sum()) == 0:
+    raise ValueError("No non-targeting controls found after control normalization.")
+
+# ---------------------------------------------------------------------
+# 4. Biological context -> obs["cell_line"].
+# ---------------------------------------------------------------------
+# Adamson is conventionally K562 in these perturbation benchmarks.
+if "cell_line" not in a.obs.columns:
+    a.obs["cell_line"] = "k562"
+else:
+    a.obs["cell_line"] = a.obs["cell_line"].astype(str)
+
+# ---------------------------------------------------------------------
+# 5. Batch/sample -> obs["gem_group"].
+# ---------------------------------------------------------------------
+# This Adamson h5ad does not expose a clear batch column in the inspection.
+# Use a dummy batch for now.
+if "gem_group" not in a.obs.columns:
+    a.obs["gem_group"] = "batch0"
+else:
+    a.obs["gem_group"] = a.obs["gem_group"].astype(str)
+
+# ---------------------------------------------------------------------
+# 6. Convert key columns to categories.
+# ---------------------------------------------------------------------
+for col in ["gene", "cell_line", "gem_group"]:
+    a.obs[col] = a.obs[col].astype(str).astype("category")
+    a.obs[col] = a.obs[col].cat.remove_unused_categories()
+
+# ---------------------------------------------------------------------
+# 7. Remove stale embeddings.
+# ---------------------------------------------------------------------
+for key in ["X_state", "X_vci"]:
+    if key in a.obsm:
+        del a.obsm[key]
+        print(f"Deleted stale obsm[{key!r}]")
+
+# ---------------------------------------------------------------------
+# 8. Create X_hvg from var["highly_variable_2000"].
+# ---------------------------------------------------------------------
+preferred_hvg_col = "highly_variable_2000"
+
+if preferred_hvg_col not in a.var.columns:
+    raise ValueError(
+        f"Expected var[{preferred_hvg_col!r}], but it is missing. "
+        "Use another HVG column or compute HVGs first."
+    )
+
+hvg_mask = a.var[preferred_hvg_col].astype(bool).to_numpy()
+print(f"\nUsing var[{preferred_hvg_col!r}] for X_hvg")
+print("HVG count:", int(hvg_mask.sum()))
+
+if int(hvg_mask.sum()) != 2000:
+    raise ValueError(f"Expected 2000 HVGs, found {int(hvg_mask.sum())}")
+
+x_hvg = a[:, hvg_mask].X
+if sp.issparse(x_hvg):
+    x_hvg = x_hvg.toarray()
+
+a.obsm["X_hvg"] = np.asarray(x_hvg, dtype=np.float32)
+print("Created obsm['X_hvg']:", a.obsm["X_hvg"].shape)
+
+# ---------------------------------------------------------------------
+# 9. Save reports.
+# ---------------------------------------------------------------------
+a.obs["gene"].astype(str).value_counts().to_csv(
+    report_dir / "patched_gene_counts.csv"
+)
+a.obs["cell_line"].astype(str).value_counts().to_csv(
+    report_dir / "patched_cell_line_counts.csv"
+)
+a.obs["gem_group"].astype(str).value_counts().to_csv(
+    report_dir / "patched_gem_group_counts.csv"
+)
+
+summary = []
+summary.append(f"source={src}")
+summary.append(f"output={dst}")
+summary.append(f"shape={a.shape}")
+summary.append(f"obs_columns={list(a.obs.columns)}")
+summary.append(f"var_columns={list(a.var.columns)}")
+summary.append(f"layers={list(a.layers.keys())}")
+summary.append(f"obsm_keys={list(a.obsm.keys())}")
+summary.append(f"gene_nunique={a.obs['gene'].astype(str).nunique()}")
+summary.append(f"cell_line_nunique={a.obs['cell_line'].astype(str).nunique()}")
+summary.append(f"gem_group_nunique={a.obs['gem_group'].astype(str).nunique()}")
+summary.append(f"non_targeting_cells={int((a.obs['gene'].astype(str) == 'non-targeting').sum())}")
+summary.append(f"X_hvg_shape={a.obsm['X_hvg'].shape}")
+
+(report_dir / "patched_schema_summary.txt").write_text(
+    "\n".join(summary) + "\n"
+)
+
+# ---------------------------------------------------------------------
+# 10. Final print and write.
+# ---------------------------------------------------------------------
+print("\nFINAL PATCHED ANNData")
+print(a)
+print("obs columns:", list(a.obs.columns))
+print("var columns:", list(a.var.columns))
+print("layers:", list(a.layers.keys()))
+print("obsm keys:", list(a.obsm.keys()))
+
+print("\ngene counts:")
+print(a.obs["gene"].astype(str).value_counts().head(30))
+
+print("\ncell_line counts:")
+print(a.obs["cell_line"].astype(str).value_counts())
+
+print("\ngem_group counts:")
+print(a.obs["gem_group"].astype(str).value_counts())
+
+print("\nWriting:", dst)
+a.write_h5ad(dst, compression="gzip")
+print("Done.")
+print("Summary:", report_dir / "patched_schema_summary.txt")

data/adamson_state_compatible/reports/patched_cell_line_counts.csv

@@ -0,0 +1,2 @@
+cell_line,count
+k562,56998

data/adamson_state_compatible/reports/patched_gem_group_counts.csv

@@ -0,0 +1,2 @@
+gem_group,count
+batch0,56998

data/adamson_state_compatible/reports/patched_gene_counts.csv

@@ -0,0 +1,78 @@
+gene,count
+non-targeting,7629
+ASCC3,1824
+SEC61B,1760
+DNAJC19,1588
+YIPF5,1457
+SCYL1,1268
+HSPA5,1242
+SEC61A1,1228
+IARS2,1142
+GBF1,1012
+HSPA9,926
+SAMM50,838
+XRN1,810
+TIMM23,789
+DAD1,774
+AMIGO3,750
+SRPRB,732
+SRP68,729
+UFM1,723
+MRPL39,712
+TELO2,711
+FECH,702
+IDH3A,690
+SEC63,684
+TTI2,682
+SYVN1,676
+DERL2,676
+SEC61G,674
+SPCS3,672
+OST4,669
+FARSB,655
+EIF2B4,654
+TMED10,649
+SLC35B1,644
+PSMD4,639
+EIF2B2,638
+GMPPB,634
+DDRGK1,633
+SLC39A7,626
+PDIA6,618
+SPCS2,612
+UFL1,611
+TIMM44,605
+MRGBP,586
+MANF,578
+P4HB,578
+PTDSS1,557
+TTI1,542
+SRP72,541
+MTHFD1,540
+TMED2,534
+EIF2S1,534
+DHDDS,526
+STT3A,523
+NEDD8,518
+GNPNAT1,514
+HYOU1,514
+ARHGAP22,513
+CHERP,497
+TMEM167A,488
+KCTD16,480
+SEL1L,464
+SOCS1,462
+EIF2B3,427
+HSD17B12,421
+CCND3,412
+DDOST,398
+ATF4,382
+CAD,295
+COPZ1,291
+PPWD1,241
+COPB1,235
+EIF2AK3,173
+ERN1,88
+XBP1,66
+PSMD12,63
+PSMA1,30

data/adamson_state_compatible/reports/patched_schema_summary.txt

@@ -0,0 +1,12 @@
+source=/projects/p32222/aTester/state_parse_repro/data/adamson_state_compatible/raw/Adamson.h5ad
+output=/projects/p32222/aTester/state_parse_repro/data/adamson_state_compatible/processed/adamson_state_schema_no_xstate.h5ad
+shape=(56998, 5043)
+obs_columns=['gene', 'perturbation', 'n_genes', 'n_genes_by_counts', 'total_counts', 'total_counts_mt', 'pct_counts_mt', 'gene_original', 'cell_line', 'gem_group']
+var_columns=['Ensembl_ID', 'n_cells', 'mt', 'n_cells_by_counts', 'mean_counts', 'pct_dropout_by_counts', 'total_counts', 'highly_variable', 'means', 'dispersions', 'dispersions_norm', 'highly_variable_500', 'highly_variable_1000', 'highly_variable_2000', 'highly_variable_5000']
+layers=['counts', 'logNor']
+obsm_keys=['X_hvg']
+gene_nunique=77
+cell_line_nunique=1
+gem_group_nunique=1
+non_targeting_cells=7629
+X_hvg_shape=(56998, 2000)

data/adamson_state_compatible/tomls/adamson_fewshot.toml

@@ -0,0 +1,9 @@
+[datasets]
+adamson = "/gpfs/projects/p32222/aTester/state_parse_repro/data/adamson_state_compatible/processed/state_tx_data"
+
+[training]
+adamson = "train"
+
+[fewshot."adamson.k562"]
+val = [ "HYOU1", "IARS2", "SPCS2", "SRP68", "TMED10",]
+test = [ "COPB1", "COPZ1", "EIF2AK3", "EIF2B4", "MTHFD1", "P4HB", "PSMD4", "SCYL1", "SRPRB", "STT3A", "UFM1",]

data/adamson_state_compatible/tomls/adamson_perturbation_counts.csv

@@ -0,0 +1,78 @@
+gene,count
+non-targeting,7629
+ASCC3,1824
+SEC61B,1760
+DNAJC19,1588
+YIPF5,1457
+SCYL1,1268
+HSPA5,1242
+SEC61A1,1228
+IARS2,1142
+GBF1,1012
+HSPA9,926
+SAMM50,838
+XRN1,810
+TIMM23,789
+DAD1,774
+AMIGO3,750
+SRPRB,732
+SRP68,729
+UFM1,723
+MRPL39,712
+TELO2,711
+FECH,702
+IDH3A,690
+SEC63,684
+TTI2,682
+SYVN1,676
+DERL2,676
+SEC61G,674
+SPCS3,672
+OST4,669
+FARSB,655
+EIF2B4,654
+TMED10,649
+SLC35B1,644
+PSMD4,639
+EIF2B2,638
+GMPPB,634
+DDRGK1,633
+SLC39A7,626
+PDIA6,618
+SPCS2,612
+UFL1,611
+TIMM44,605
+MRGBP,586
+MANF,578
+P4HB,578
+PTDSS1,557
+TTI1,542
+SRP72,541
+MTHFD1,540
+TMED2,534
+EIF2S1,534
+DHDDS,526
+STT3A,523
+NEDD8,518
+GNPNAT1,514
+HYOU1,514
+ARHGAP22,513
+CHERP,497
+TMEM167A,488
+KCTD16,480
+SEL1L,464
+SOCS1,462
+EIF2B3,427
+HSD17B12,421
+CCND3,412
+DDOST,398
+ATF4,382
+CAD,295
+COPZ1,291
+PPWD1,241
+COPB1,235
+EIF2AK3,173
+ERN1,88
+XBP1,66
+PSMD12,63
+PSMA1,30

data/adamson_state_compatible/tomls/adamson_split_summary.csv

@@ -0,0 +1,14 @@
+,0
+h5ad,/projects/p32222/aTester/state_parse_repro/data/adamson_state_compatible/processed/state_tx_data/adamson.h5ad
+data_dir,/gpfs/projects/p32222/aTester/state_parse_repro/data/adamson_state_compatible/processed/state_tx_data
+toml,/projects/p32222/aTester/state_parse_repro/data/adamson_state_compatible/tomls/adamson_fewshot.toml
+cell_line,k562
+control,non-targeting
+min_cells,50
+eligible_perturbations,75
+val_perturbations,5
+test_perturbations,11
+train_like_perturbations,59
+total_cells,56998
+total_genes,5043
+non_targeting_cells,7629

data/replogle_nadig_hf/replogle.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c53aadd1bbbaee7604b579b2592034a0b9d9bfa9eecfc6cdc622a9d2e205c77
+size 22297813040

data/replogle_nadig_hf/rpe1_normalized_singlecell_01.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91282e58371ae9d98a3283af0d69fc9e3b0984386437139e7d44dba2c60f232e
+size 8701143594

data/replogle_nogwps_v2/rpe1.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e0679a79597a648993942bae4229e807b6e158201db692bd7e74aea8b2490c4
+size 2581402168

data/replogle_nogwps_v2_rpe1_checkpoint_vocab/rpe1.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49193df3689d5e92450bde5dd507b3d1783de5204f4d357176bd8f59cfb32464
+size 2081026707

data/replogle_self_se/rpe1_raw_from_replogle.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c708ae8c0e23cdbc85aa1c57ba3c1980fd8e3c6c57bebea4ece8fca1f90b8d5b
+size 1513439710

data/replogle_self_se/rpe1_self_se_embedded.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff3dff7b54077dbfdc4e13e4a8f4cf8d1b33a6058c89e430ec7beeb2491d3219
+size 2803013789

data/replogle_self_se_for_st/rpe1.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3305969590216fecc89d6e5efa93ff0ba73f83da432f8899547663a1988727a1
+size 2803033451

data/replogle_source_filtered_schema_for_st/replogle_concat.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7363e01c57adccf90244044c959efdb1e5fdbda4778d62a6e0b2199c6c21ad83
+size 27342995251

data/state_replogle_filtered/all.toml

@@ -0,0 +1,16 @@
+[datasets]
+replogle = "/data/replogle_nogwps_v2" # ADDS ALL h5 or h5ad files in this folder to training
+
+# Training specifications
+# All cell types in a dataset automatically go into training (excluding zeroshot/fewshot overrides)
+[training]
+replogle = "train"
+
+# Zeroshot specifications - entire cell types go to val or test
+[zeroshot]
+
+# Fewshot specifications - explicit perturbation lists
+[fewshot]
+[fewshot."replogle.hepg2"]
+val = []
+test = []

data/state_replogle_filtered/replogle_concat.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7992a6109d589f2d49c2eaccff45a9f860418af23125d41b80d8e3929a33d56
+size 30010775846

data/state_replogle_filtered/rpe1.toml

@@ -0,0 +1,16 @@
+[datasets]
+replogle = "/data/replogle_nogwps_v2" # ADDS ALL h5 or h5ad files in this folder to training
+
+# Training specifications
+# All cell types in a dataset automatically go into training (excluding zeroshot/fewshot overrides)
+[training]
+replogle = "train"
+
+# Zeroshot specifications - entire cell types go to val or test
+[zeroshot]
+
+# Fewshot specifications - explicit perturbation lists
+[fewshot]
+[fewshot."replogle.rpe1"]
+test = ['RPA3', 'POLR1E', 'TMEM258', 'PPIH', 'SMC2', 'NUBP2', 'RNF8', 'ATF4', 'USPL1', 'SDHA', 'RFT1', 'IPO7', 'PSMG2', 'SRSF10', 'MRPL23', 'ZC3H4', 'STRAP', 'SBDS', 'COPS8', 'LSM12', 'GTF2F1', 'PPP2R3C', 'VPS29', 'USP9X', 'TRAPPC8', 'PFDN4', 'METTL17', 'CNOT9', 'VPS35', 'FNBP4', 'XRCC1', 'CEBPZ', 'ATP5F1B', 'MAX', 'POLA1', 'CYREN', 'REXO2', 'PSMA4', 'CSTF1', 'MRPS17', 'RPL36A', 'MRPL55', 'ZNF143', 'CEP152', 'SSU72', 'MRPL45', 'CDT1', 'SRF', 'RPL23A', 'RBX1', 'SLU7', 'WDR12', 'NAGLU', 'LSM5', 'EFTUD2', 'PALB2', 'BCR', 'NUP88', 'MIPEP', 'UTP23', 'NUP107', 'DYNC1H1', 'DYNLL2', 'POLR3C', 'KDM1A', 'TNRC6A', 'NMD3', 'RFC4', 'SMG5', 'PSMD14', 'NUP160', 'POLR2I', 'HMGN2', 'RPL17', 'SLBP', 'IARS2', 'VBP1', 'ZRANB2', 'GPS2', 'DAXX', 'CDK9', 'ELP3', 'NSL1', 'MANF', 'TBPL1', 'UBE2H', 'TIMM8A', 'POP5', 'PDCD7', 'RAB4B', 'NARS', 'TMEM214', 'TUBB2A', 'ANAPC13', 'SBNO1', 'RMI1', 'ALG2', 'NDUFB3', 'GEMIN6', 'MTHFD1', 'RFC2', 'DDOST', 'SCNM1', 'WDR1', 'PPP1R15B', 'CHTOP', 'MRPS31', 'COQ4', 'SP2', 'TOP1', 'CLPB', 'ATP6V1A', 'ORC1', 'GFM1', 'NDC1', 'GABPB1', 'HTATSF1', 'LSM10', 'TADA3', 'SPEN', 'PYROXD1', 'SF3B5', 'ESYT2', 'RPAP1', 'PSMA3', 'RPL30', 'CNOT11', 'TIMM44', 'MRPL44', 'NELFA', 'ENY2', 'LETM1', 'CNOT3', 'OXA1L', 'CCDC59', 'MMS19', 'NKAP', 'BANP', 'KCTD10', 'CMTR1', 'SMAGP', 'RAD17', 'URM1', 'GFER', 'PAF1', 'PARN', 'EIF2B2', 'NSMCE2', 'GPN3', 'DHDDS', 'SETD5', 'BCL2L1', 'ALG1', 'ANAPC11', 'GLE1', 'NCKAP1', 'XRCC2', 'DCTN6', 'LSG1', 'ATP6V1G1', 'C1D', 'TRNT1', 'TRAPPC5', 'FAF2', 'POLR2B', 'NEMF', 'PRELID1', 'E2F6', 'AP2S1', 'RPN1', 'CAMLG', 'NOC3L', 'BAP1', 'CWC25', 'PHB', 'HNRNPK', 'SDHAF2', 'SNTB2', 'PRPF19', 'TUBGCP3', 'GPS1', 'ZCRB1', 'NUP54', 'KTI12', 'EXOSC3', 'TERF1', 'KATNB1', 'MRPS6', 'SNRPD3', 'CDC16', 'ILF2', 'TBP', 'COX6C', 'ZRSR2', 'ANKRD39', 'MRPL27', 'ATP6V1D', 'RPL6', 'ELOB', 'INO80C', 'TUBGCP5', 'MRPL39', 'TADA2B', 'YTHDC1', 'TCERG1', 'THAP1', 'ACTR6', 'RPIA', 'SGPP1', 'MRPL36', 'MRPL4', 'SPATA5', 'MPG', 'CCDC115', 'GEMIN5', 'PNN', 'MRPS33', 'UPF1', 'BAG6', 'ORC6', 'BET1', 'BRK1', 'TRMT10C', 'GTF3A', 'SRSF6', 'MARS', 'BTF3', 'TAF6', 'TARS2', 'KXD1', 'FAM32A', 'NAE1', 'MAD2L2', 'NCBP2', 'NUBP1', 'PABPN1', 'KIF4A', 'LRP5', 'ABCE1', 'TSR2', 'CNOT2', 'CENPH', 'INTS1', 'THAP11', 'MSRB1', 'MRPS35', 'MRPL17', 'JMJD6', 'MRPS10', 'WDR54', 'YEATS2', 'PSMD13', 'POP4', 'SYF2', 'HUS1', 'DPM2', 'SNRPC', 'USP10', 'CSE1L', 'ERAL1', 'TBCD', 'RPL8', 'SNRPD1', 'GCLC', 'PRMT1', 'DAP3', 'KRI1', 'NELFB', 'MRPL41', 'MRPL46', 'GTF3C2', 'GEMIN8', 'SHQ1', 'PFN1', 'FAU', 'MRPS16', 'DBR1', 'HLA-C', 'CNN2', 'TAF3', 'INTS2', 'ELP2', 'COPS6', 'NAF1', 'BCLAF1', 'LSM2', 'PGAM5', 'SPCS3', 'METAP1', 'MAD2L1', 'TOMM40', 'ABHD11', 'PGAM1', 'TMEM161B', 'WTAP', 'CRLS1', 'DCTN1', 'H2AFZ', 'EHMT2', 'RPL28', 'MCM6', 'HSPA9', 'ATP11B', 'ILF3', 'RRS1', 'CARS', 'SMC1A', 'CHMP5', 'PHF10', 'MRPS11', 'MMGT1', 'PTPN23', 'NAA30', 'TRAPPC11', 'TP53RK', 'ZDHHC7', 'CD2BP2', 'NDUFAF3', 'COQ2', 'NCOA4', 'TEN1', 'SNRPE', 'GTF3C4', 'HYOU1', 'SHC1', 'PKMYT1', 'HSP90B1', 'POGLUT3', 'KPNA6', 'C5orf30', 'RPP21', 'COX5A', 'NIFK', 'UBR5', 'CCND3', 'DAD1', 'MRPS18A', 'MNAT1', 'NUP133', 'KLC2', 'GSPT1', 'EXOC5', 'TPR', 'PLK4', 'DDX11', 'DTL', 'MRPL38', 'MRPS18C', 'PSMC1', 'ELOF1', 'KIAA0586', 'DNAJC11', 'ATP6AP1', 'CPNE1', 'DDX52', 'DDX20', 'GRWD1', 'PSMB5', 'DDX27', 'TEFM', 'CSDE1', 'CXXC1', 'HAUS1', 'TAF1D', 'RARS2', 'NCAPH2', 'ATP6AP2', 'TRPM7', 'ATXN10', 'SLC7A6OS', 'TOE1', 'NRBP1', 'EXOC3', 'GTF2H4', 'TRA2B', 'UPF2', 'ANAPC5', 'GUK1', 'RCOR1', 'PRPF8', 'LAMTOR4', 'RAB6A', 'EIF2B5', 'TTF2', 'KIN', 'RBBP4', 'VEZT', 'PPP1R37', 'PPP6C', 'SARNP', 'ARPC4', 'GEMIN4', 'RPLP0', 'TSG101', 'TAF2', 'EXOSC7', 'EIF3J', 'MBTPS1', 'EIF4B', 'SKA3', 'ARGLU1', 'GFOD2', 'TIPIN', 'PSAT1', 'FEN1', 'STX5', 'TUBGCP2', 'RPL14', 'CCNB1', 'DHX16', 'WDR77', 'KDM5C', 'GTF3C3', 'VHL', 'WDR43', 'RANGAP1', 'ALG13', 'ZNHIT2', 'NCAPH', 'TUBB', 'TACC3', 'TFAM', 'SDC1', 'VPS13D', 'EIF2B4', 'UXS1', 'BRCA2', 'SKP2', 'PABPC4', 'COX17', 'EEF1B2', 'PRKRA', 'BTF3L4', 'PRPF40A', 'SEC61B', 'RBMXL1', 'CDIPT', 'RPLP2', 'RPN2', 'SETDB1', 'COG4', 'EDC4', 'SNRPA', 'RABGGTA', 'CDK2', 'ATP5PO', 'SP1', 'GLRX5', 'MRPL35', 'MRPS24', 'FAM207A', 'METTL16', 'EIF1AX', 'NARS2', 'MRPS9', 'SRP68', 'ADSL', 'MCMBP', 'SEM1', 'TARDBP', 'DDX19A', 'ASF1B', 'RPE', 'TOMM22', 'HDAC7', 'SAMM50', 'MBTPS2', 'TSEN54', 'EIF3D', 'MRPS25', 'NSA2', 'PIGH', 'RNASEH2C', 'CNOT10', 'ATL2', 'SF3B3', 'WAC', 'SCFD1', 'OGFOD1', 'RPL13', 'SHOC2', 'EXOC1', 'HAUS8', 'KIF14', 'ACTR8', 'TPT1', 'PPP1R12A', 'RUVBL1', 'DDX23', 'SMNDC1', 'MRPL11', 'IMPDH2', 'CSTF3', 'LRR1', 'POLG', 'RTTN', 'IGBP1', 'RPA2', 'RPL29', 'COX10', 'RPS13', 'RPUSD3', 'NFYC', 'DDX41', 'PMPCA', 'ZFR', 'ATP6V1F', 'MRPL24', 'RSRC2', 'AAAS', 'ATP6V0B', 'ATP6V1E1', 'SLC35G2', 'SS18L2', 'PSMC5', 'PCBP1', 'HSPA14', 'SRSF7', 'PSMB3', 'TIMM22', 'TFRC', 'HARS', 'REV3L', 'EIF2S3', 'HNRNPR', 'VPS54', 'CAP1', 'CENPW', 'DGCR8', 'PTPN11', 'VPS18', 'UQCRQ', 'DHPS', 'RBM33', 'KAT7', 'GTF3C5', 'MRPL51', 'CAPZB', 'EXOC2', 'CENPJ', 'INO80', 'PRRC2A', 'IARS', 'TAZ', 'TXNL4B', 'SLC35B1', 'UNC50', 'VARS', 'ZNF720', 'PRIM1', 'RBM10', 'MRPL42', 'SNRNP200', 'CHMP6', 'EIF5A', 'NASP', 'EIF3B', 'CNOT1', 'POGZ', 'MBD3', 'SPRTN', 'MIOS', 'GTF2H3', 'TAF11', 'CENPK', 'DRG1', 'IPO9', 'H2AFX', 'ACD', 'SRA1', 'NAA35', 'CHORDC1', 'PPRC1', 'EPRS', 'WDR18', 'PPP4R2', 'XRCC5', 'ENO1', 'TMEM242', 'HECTD1', 'PGD', 'PGS1', 'PNKP', 'MRPS30', 'ZFC3H1', 'ECD', 'EIF4G2', 'RPS8', 'EIF2B1', 'EIF2S1', 'NPLOC4', 'ATP6V0D1', 'MOB4', 'HNRNPC', 'HUWE1', 'NDUFA3', 'LRPPRC', 'NHLRC2', 'MRGBP', 'ATP1A1', 'BMS1', 'DRAP1', 'GRSF1', 'SUDS3', 'MTG2', 'MTPAP', 'PMF1', 'SEH1L', 'CASC3', 'HIPK1', 'CALR', 'PIAS1', 'ZNF292', 'ATR', 'RIOK1', 'THRAP3', 'CNIH4', 'UBR4', 'CHD4', 'THOC6', 'CTNNBL1', 'CSNK2B', 'MED14', 'PXN', 'CCNC', 'BANF1', 'SRP72', 'NUTF2', 'KIAA1143', 'WDHD1', 'NMT1', 'BRIP1', 'TFB1M', 'NOP14', 'MFN2', 'EXOSC2', 'LARS', 'SART1', 'DNAJA3', 'SMC6', 'SLC7A5', 'ORC5', 'RPS26', 'ATIC', 'MRPL18', 'CTR9', 'TEAD3', 'PRXL2A', 'HMGA1', 'LSM4', 'RBBP5', 'RPAP3', 'CHCHD3', 'EIF4G1', 'TELO2', 'LONP1', 'UBL5', 'GOLT1B', 'LIMS1', 'TEX10', 'GARS', 'GNL3', 'SLC25A3', 'RBM42', 'RPRD1B', 'CCT7', 'DENR', 'CYCS', 'ACTR1B', 'NLE1', 'ACTB', 'MRPL20', 'DCUN1D5', 'WARS', 'RTCB', 'TP53I13', 'BRF1', 'NAA25', 'RPL23', 'SUPT20H', 'UBA5', 'MCM3', 'NUP62', 'TSEN2', 'NDUFV2', 'UQCRH', 'SERBP1', 'EIF3A', 'ZMAT5', 'ATP6V1C1', 'AIFM1', 'PDCD11', 'TMED2', 'NSUN4', 'DUT', 'CRCP', 'MRPS23', 'CMPK1', 'SNAPC1', 'MIS18BP1', 'SPTLC1', 'RNF168', 'ZW10', 'ETF1', 'RPL26L1', 'IPO13', 'ARL2', 'RPL10A', 'DPH2', 'MRPL1', 'CENPN', 'TOP3A', 'DSN1', 'MCM4', 'VPS37A', 'TUFM', 'RCL1', 'STIL', 'RPL24', 'RBM25', 'GBF1', 'DNAJC17', 'NISCH', 'CLOCK', 'TUBA1B', 'GTF2H2', 'GRPEL1', 'PPIL1', 'SMG7', 'UBQLN4', 'RBM4', 'DBF4', 'PPP2CA', 'MIS18A', 'YKT6', 'SDHC', 'TWF1', 'TOX4', 'RPS27A', 'STRIP1', 'AURKA', 'DYNC1I2', 'SUPV3L1', 'EGLN2', 'OIP5', 'NAA20', 'CINP', 'POLD3', 'OSTC', 'ITGB1BP1', 'PPP2CB', 'BUB1', 'ZNHIT1', 'DIMT1', 'HEXIM1', 'TSFM', 'POLR2M', 'RPL35', 'RAD51D', 'HSPE1', 'COA5', 'CPNE7', 'FDXR', 'STXBP4', 'GINS4', 'PAM16', 'MCM2', 'FKBP9', 'GYG1', 'TTK', 'RPS27', 'ANKS6', 'PPIL2', 'TTC4', 'MRPL14', 'TMX2', 'LAMTOR1', 'RUVBL2', 'ALG14', 'KEAP1', 'CPOX', 'TOMM20', 'CCDC137', 'MRPS21', 'TUBA1C', 'TRMT5', 'CDC73', 'BUD13', 'DNAJA1', 'SMARCE1', 'DMAP1', 'SNUPN', 'NAPG', 'CENPE', 'INO80B', 'RTRAF', 'SUPT16H', 'CPSF3', 'CUL1', 'CUL3', 'CCNH', 'EIF4H', 'PPME1', 'ZNF787', 'RAC3', 'MSL1', 'ATP6V1B2', 'SPG7', 'DPY19L4', 'SPC24', 'OPA1', 'CABIN1', 'POLL', 'SNRPB', 'PRPF39', 'NSMCE1', 'RPS19', 'SLC39A10', 'TMEM199', 'YARS', 'EIF4E', 'CCT4', 'DYNLL1', 'SAE1', 'ANAPC1', 'MRPS34', 'HNRNPU', 'COX7C', 'NUP85', 'METTL1', 'UROD', 'SUMO2', 'WDR44', 'IMPA2', 'IST1', 'MRPL32', 'PEF1', 'RNF14', 'TTI1', 'EIF2S2', 'PPIL4', 'ZBTB17', 'RNPC3', 'TBCB', 'GPN2', 'VPS41', 'SRPRB', 'COQ5', 'POLD1', 'WDR33', 'UMPS', 'TUBG1', 'METTL23', 'CCP110', 'MRPL49', 'CDC6', 'ARL4D', 'CS', 'RPA1', 'HMGCS1', 'NUB1', 'GMPPB', 'SNRNP27', 'NCAPD2', 'CHCHD2', 'NFYB', 'RNF20', 'EIF1', 'PTK2', 'ZNF236', 'VPS28', 'TFDP1', 'C9orf16', 'PDCD6', 'GART', 'PSME1', 'YEATS4', 'U2SURP', 'TNPO3', 'CENPC', 'ERCC2', 'ATP5MF', 'POLD2', 'CLTC', 'EEF1G', 'ISCA2', 'RNMT', 'DNAJC19', 'NSF', 'CHMP7', 'DHX33', 'SARS2', 'DSTYK', 'DPAGT1', 'UBE2M', 'MRPL22', 'CENPT', 'NDUFA8', 'INTS13', 'UQCRC2', 'GPKOW', 'ATF5', 'BRIX1', 'SASS6', 'ACTR2', 'HAUS4', 'CTU2', 'POLRMT', 'MRPS7', 'MMS22L', 'EIF3L', 'WDR4', 'TMEM127', 'ELP6', 'RPL15', 'GSK3B', 'MYBL2', 'WBP1', 'MRPL37', 'GTF3C6', 'AP2M1', 'DDX17', 'SET', 'EIF3M', 'PRORP', 'RNGTT', 'MFAP1', 'HJURP', 'MRPS14', 'EPB41L2', 'WDR5', 'BUB1B', 'RPL32', 'TRIAP1', 'FASTKD5', 'CCDC84', 'COG1', 'FDPS', 'DNAJC8', 'FUNDC2', 'EXOC7', 'NUDCD3', 'XRN1', 'MBIP', 'YARS2', 'DDX55', 'URI1', 'NDUFB10', 'EIF3I', 'TUBGCP4', 'SLC2A8', 'TTC1', 'MPHOSPH6', 'UFM1', 'UBE2I', 'CBLL1', 'PITRM1', 'NUP153', 'VPS33A', 'MRPS26', 'EIF2B3', 'CYC1', 'CEP97', 'SEPHS2', 'PLEKHN1', 'EXOSC1', 'INTS14', 'MCM5', 'AGBL5', 'WDR83OS', 'ARMC6', 'AFG3L2', 'RHOQ', 'HBS1L', 'RPL35A', 'PDCD5', 'AASDHPPT', 'FAM136A', 'HPS5', 'HDAC3', 'YPEL5', 'RPS2', 'SEC63', 'NDUFA11', 'SRRT', 'ADAM10', 'GET3', 'TUBE1', 'EIF6', 'NUP214', 'NDUFB8', 'RPL5', 'PET117', 'NAA50', 'SPCS2', 'SNAPC3', 'ELL', 'PRIM2', 'MRPL34', 'TLCD1', 'CCT6A', 'TIMM10', 'RPL26', 'MVD', 'HMGB3', 'CIAO1', 'ADRM1', 'HNRNPA1', 'SNRNP25', 'FXN', 'MTOR', 'IWS1', 'SSBP3', 'BDP1', 'COMTD1', 'MICOS10', 'RPL39', 'HSD17B10', 'SMG8', 'DPH6', 'DCTN4', 'PTEN', 'PHB2', 'MED11', 'RAB18', 'PES1', 'NKAPD1', 'NUDC', 'RPS3', 'CDK1', 'COG6', 'EXOSC4', 'LAS1L', 'RPP14', 'EXOSC8', 'RPS6', 'NOM1', 'DDX1', 'FBXW7', 'KCMF1', 'PTPN1', 'TXN', 'RPL9', 'MRPL19', 'MRPL28', 'GNPNAT1', 'PPA2', 'PGK1', 'SNRPG', 'SAMD4B', 'NOLC1', 'EEFSEC', 'AHCTF1', 'CLNS1A', 'TAF8', 'DOHH', 'NUP35', 'RPUSD4', 'EP400', 'IFITM2', 'THOC7']
+# train gets all other perturbations automatically

data/state_replogle_filtered/rpe1_zeroshot.toml

@@ -0,0 +1,14 @@
+[datasets]
+replogle = "/data/replogle_nogwps_v2" # ADDS ALL h5 or h5ad files in this folder to training
+
+# Training specifications
+# All cell types in a dataset automatically go into training (excluding zeroshot/fewshot overrides)
+[training]
+replogle = "train"
+
+# Zeroshot specifications - entire cell types go to val or test
+[zeroshot]
+"replogle.rpe1" = "test"
+
+# Fewshot specifications - explicit perturbation lists
+[fewshot]

data/state_replogle_filtered_analysis/all.toml

@@ -0,0 +1,16 @@
+[datasets]
+replogle = "/data/replogle_nogwps_v2" # ADDS ALL h5 or h5ad files in this folder to training
+
+# Training specifications
+# All cell types in a dataset automatically go into training (excluding zeroshot/fewshot overrides)
+[training]
+replogle = "train"
+
+# Zeroshot specifications - entire cell types go to val or test
+[zeroshot]
+
+# Fewshot specifications - explicit perturbation lists
+[fewshot]
+[fewshot."replogle.hepg2"]
+val = []
+test = []

data/state_replogle_filtered_analysis/replogle_concat.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7992a6109d589f2d49c2eaccff45a9f860418af23125d41b80d8e3929a33d56
+size 30010775846

data/state_replogle_filtered_analysis/rpe1.toml

@@ -0,0 +1,16 @@
+[datasets]
+replogle = "/data/replogle_nogwps_v2" # ADDS ALL h5 or h5ad files in this folder to training
+
+# Training specifications
+# All cell types in a dataset automatically go into training (excluding zeroshot/fewshot overrides)
+[training]
+replogle = "train"
+
+# Zeroshot specifications - entire cell types go to val or test
+[zeroshot]
+
+# Fewshot specifications - explicit perturbation lists
+[fewshot]
+[fewshot."replogle.rpe1"]
+test = ['RPA3', 'POLR1E', 'TMEM258', 'PPIH', 'SMC2', 'NUBP2', 'RNF8', 'ATF4', 'USPL1', 'SDHA', 'RFT1', 'IPO7', 'PSMG2', 'SRSF10', 'MRPL23', 'ZC3H4', 'STRAP', 'SBDS', 'COPS8', 'LSM12', 'GTF2F1', 'PPP2R3C', 'VPS29', 'USP9X', 'TRAPPC8', 'PFDN4', 'METTL17', 'CNOT9', 'VPS35', 'FNBP4', 'XRCC1', 'CEBPZ', 'ATP5F1B', 'MAX', 'POLA1', 'CYREN', 'REXO2', 'PSMA4', 'CSTF1', 'MRPS17', 'RPL36A', 'MRPL55', 'ZNF143', 'CEP152', 'SSU72', 'MRPL45', 'CDT1', 'SRF', 'RPL23A', 'RBX1', 'SLU7', 'WDR12', 'NAGLU', 'LSM5', 'EFTUD2', 'PALB2', 'BCR', 'NUP88', 'MIPEP', 'UTP23', 'NUP107', 'DYNC1H1', 'DYNLL2', 'POLR3C', 'KDM1A', 'TNRC6A', 'NMD3', 'RFC4', 'SMG5', 'PSMD14', 'NUP160', 'POLR2I', 'HMGN2', 'RPL17', 'SLBP', 'IARS2', 'VBP1', 'ZRANB2', 'GPS2', 'DAXX', 'CDK9', 'ELP3', 'NSL1', 'MANF', 'TBPL1', 'UBE2H', 'TIMM8A', 'POP5', 'PDCD7', 'RAB4B', 'NARS', 'TMEM214', 'TUBB2A', 'ANAPC13', 'SBNO1', 'RMI1', 'ALG2', 'NDUFB3', 'GEMIN6', 'MTHFD1', 'RFC2', 'DDOST', 'SCNM1', 'WDR1', 'PPP1R15B', 'CHTOP', 'MRPS31', 'COQ4', 'SP2', 'TOP1', 'CLPB', 'ATP6V1A', 'ORC1', 'GFM1', 'NDC1', 'GABPB1', 'HTATSF1', 'LSM10', 'TADA3', 'SPEN', 'PYROXD1', 'SF3B5', 'ESYT2', 'RPAP1', 'PSMA3', 'RPL30', 'CNOT11', 'TIMM44', 'MRPL44', 'NELFA', 'ENY2', 'LETM1', 'CNOT3', 'OXA1L', 'CCDC59', 'MMS19', 'NKAP', 'BANP', 'KCTD10', 'CMTR1', 'SMAGP', 'RAD17', 'URM1', 'GFER', 'PAF1', 'PARN', 'EIF2B2', 'NSMCE2', 'GPN3', 'DHDDS', 'SETD5', 'BCL2L1', 'ALG1', 'ANAPC11', 'GLE1', 'NCKAP1', 'XRCC2', 'DCTN6', 'LSG1', 'ATP6V1G1', 'C1D', 'TRNT1', 'TRAPPC5', 'FAF2', 'POLR2B', 'NEMF', 'PRELID1', 'E2F6', 'AP2S1', 'RPN1', 'CAMLG', 'NOC3L', 'BAP1', 'CWC25', 'PHB', 'HNRNPK', 'SDHAF2', 'SNTB2', 'PRPF19', 'TUBGCP3', 'GPS1', 'ZCRB1', 'NUP54', 'KTI12', 'EXOSC3', 'TERF1', 'KATNB1', 'MRPS6', 'SNRPD3', 'CDC16', 'ILF2', 'TBP', 'COX6C', 'ZRSR2', 'ANKRD39', 'MRPL27', 'ATP6V1D', 'RPL6', 'ELOB', 'INO80C', 'TUBGCP5', 'MRPL39', 'TADA2B', 'YTHDC1', 'TCERG1', 'THAP1', 'ACTR6', 'RPIA', 'SGPP1', 'MRPL36', 'MRPL4', 'SPATA5', 'MPG', 'CCDC115', 'GEMIN5', 'PNN', 'MRPS33', 'UPF1', 'BAG6', 'ORC6', 'BET1', 'BRK1', 'TRMT10C', 'GTF3A', 'SRSF6', 'MARS', 'BTF3', 'TAF6', 'TARS2', 'KXD1', 'FAM32A', 'NAE1', 'MAD2L2', 'NCBP2', 'NUBP1', 'PABPN1', 'KIF4A', 'LRP5', 'ABCE1', 'TSR2', 'CNOT2', 'CENPH', 'INTS1', 'THAP11', 'MSRB1', 'MRPS35', 'MRPL17', 'JMJD6', 'MRPS10', 'WDR54', 'YEATS2', 'PSMD13', 'POP4', 'SYF2', 'HUS1', 'DPM2', 'SNRPC', 'USP10', 'CSE1L', 'ERAL1', 'TBCD', 'RPL8', 'SNRPD1', 'GCLC', 'PRMT1', 'DAP3', 'KRI1', 'NELFB', 'MRPL41', 'MRPL46', 'GTF3C2', 'GEMIN8', 'SHQ1', 'PFN1', 'FAU', 'MRPS16', 'DBR1', 'HLA-C', 'CNN2', 'TAF3', 'INTS2', 'ELP2', 'COPS6', 'NAF1', 'BCLAF1', 'LSM2', 'PGAM5', 'SPCS3', 'METAP1', 'MAD2L1', 'TOMM40', 'ABHD11', 'PGAM1', 'TMEM161B', 'WTAP', 'CRLS1', 'DCTN1', 'H2AFZ', 'EHMT2', 'RPL28', 'MCM6', 'HSPA9', 'ATP11B', 'ILF3', 'RRS1', 'CARS', 'SMC1A', 'CHMP5', 'PHF10', 'MRPS11', 'MMGT1', 'PTPN23', 'NAA30', 'TRAPPC11', 'TP53RK', 'ZDHHC7', 'CD2BP2', 'NDUFAF3', 'COQ2', 'NCOA4', 'TEN1', 'SNRPE', 'GTF3C4', 'HYOU1', 'SHC1', 'PKMYT1', 'HSP90B1', 'POGLUT3', 'KPNA6', 'C5orf30', 'RPP21', 'COX5A', 'NIFK', 'UBR5', 'CCND3', 'DAD1', 'MRPS18A', 'MNAT1', 'NUP133', 'KLC2', 'GSPT1', 'EXOC5', 'TPR', 'PLK4', 'DDX11', 'DTL', 'MRPL38', 'MRPS18C', 'PSMC1', 'ELOF1', 'KIAA0586', 'DNAJC11', 'ATP6AP1', 'CPNE1', 'DDX52', 'DDX20', 'GRWD1', 'PSMB5', 'DDX27', 'TEFM', 'CSDE1', 'CXXC1', 'HAUS1', 'TAF1D', 'RARS2', 'NCAPH2', 'ATP6AP2', 'TRPM7', 'ATXN10', 'SLC7A6OS', 'TOE1', 'NRBP1', 'EXOC3', 'GTF2H4', 'TRA2B', 'UPF2', 'ANAPC5', 'GUK1', 'RCOR1', 'PRPF8', 'LAMTOR4', 'RAB6A', 'EIF2B5', 'TTF2', 'KIN', 'RBBP4', 'VEZT', 'PPP1R37', 'PPP6C', 'SARNP', 'ARPC4', 'GEMIN4', 'RPLP0', 'TSG101', 'TAF2', 'EXOSC7', 'EIF3J', 'MBTPS1', 'EIF4B', 'SKA3', 'ARGLU1', 'GFOD2', 'TIPIN', 'PSAT1', 'FEN1', 'STX5', 'TUBGCP2', 'RPL14', 'CCNB1', 'DHX16', 'WDR77', 'KDM5C', 'GTF3C3', 'VHL', 'WDR43', 'RANGAP1', 'ALG13', 'ZNHIT2', 'NCAPH', 'TUBB', 'TACC3', 'TFAM', 'SDC1', 'VPS13D', 'EIF2B4', 'UXS1', 'BRCA2', 'SKP2', 'PABPC4', 'COX17', 'EEF1B2', 'PRKRA', 'BTF3L4', 'PRPF40A', 'SEC61B', 'RBMXL1', 'CDIPT', 'RPLP2', 'RPN2', 'SETDB1', 'COG4', 'EDC4', 'SNRPA', 'RABGGTA', 'CDK2', 'ATP5PO', 'SP1', 'GLRX5', 'MRPL35', 'MRPS24', 'FAM207A', 'METTL16', 'EIF1AX', 'NARS2', 'MRPS9', 'SRP68', 'ADSL', 'MCMBP', 'SEM1', 'TARDBP', 'DDX19A', 'ASF1B', 'RPE', 'TOMM22', 'HDAC7', 'SAMM50', 'MBTPS2', 'TSEN54', 'EIF3D', 'MRPS25', 'NSA2', 'PIGH', 'RNASEH2C', 'CNOT10', 'ATL2', 'SF3B3', 'WAC', 'SCFD1', 'OGFOD1', 'RPL13', 'SHOC2', 'EXOC1', 'HAUS8', 'KIF14', 'ACTR8', 'TPT1', 'PPP1R12A', 'RUVBL1', 'DDX23', 'SMNDC1', 'MRPL11', 'IMPDH2', 'CSTF3', 'LRR1', 'POLG', 'RTTN', 'IGBP1', 'RPA2', 'RPL29', 'COX10', 'RPS13', 'RPUSD3', 'NFYC', 'DDX41', 'PMPCA', 'ZFR', 'ATP6V1F', 'MRPL24', 'RSRC2', 'AAAS', 'ATP6V0B', 'ATP6V1E1', 'SLC35G2', 'SS18L2', 'PSMC5', 'PCBP1', 'HSPA14', 'SRSF7', 'PSMB3', 'TIMM22', 'TFRC', 'HARS', 'REV3L', 'EIF2S3', 'HNRNPR', 'VPS54', 'CAP1', 'CENPW', 'DGCR8', 'PTPN11', 'VPS18', 'UQCRQ', 'DHPS', 'RBM33', 'KAT7', 'GTF3C5', 'MRPL51', 'CAPZB', 'EXOC2', 'CENPJ', 'INO80', 'PRRC2A', 'IARS', 'TAZ', 'TXNL4B', 'SLC35B1', 'UNC50', 'VARS', 'ZNF720', 'PRIM1', 'RBM10', 'MRPL42', 'SNRNP200', 'CHMP6', 'EIF5A', 'NASP', 'EIF3B', 'CNOT1', 'POGZ', 'MBD3', 'SPRTN', 'MIOS', 'GTF2H3', 'TAF11', 'CENPK', 'DRG1', 'IPO9', 'H2AFX', 'ACD', 'SRA1', 'NAA35', 'CHORDC1', 'PPRC1', 'EPRS', 'WDR18', 'PPP4R2', 'XRCC5', 'ENO1', 'TMEM242', 'HECTD1', 'PGD', 'PGS1', 'PNKP', 'MRPS30', 'ZFC3H1', 'ECD', 'EIF4G2', 'RPS8', 'EIF2B1', 'EIF2S1', 'NPLOC4', 'ATP6V0D1', 'MOB4', 'HNRNPC', 'HUWE1', 'NDUFA3', 'LRPPRC', 'NHLRC2', 'MRGBP', 'ATP1A1', 'BMS1', 'DRAP1', 'GRSF1', 'SUDS3', 'MTG2', 'MTPAP', 'PMF1', 'SEH1L', 'CASC3', 'HIPK1', 'CALR', 'PIAS1', 'ZNF292', 'ATR', 'RIOK1', 'THRAP3', 'CNIH4', 'UBR4', 'CHD4', 'THOC6', 'CTNNBL1', 'CSNK2B', 'MED14', 'PXN', 'CCNC', 'BANF1', 'SRP72', 'NUTF2', 'KIAA1143', 'WDHD1', 'NMT1', 'BRIP1', 'TFB1M', 'NOP14', 'MFN2', 'EXOSC2', 'LARS', 'SART1', 'DNAJA3', 'SMC6', 'SLC7A5', 'ORC5', 'RPS26', 'ATIC', 'MRPL18', 'CTR9', 'TEAD3', 'PRXL2A', 'HMGA1', 'LSM4', 'RBBP5', 'RPAP3', 'CHCHD3', 'EIF4G1', 'TELO2', 'LONP1', 'UBL5', 'GOLT1B', 'LIMS1', 'TEX10', 'GARS', 'GNL3', 'SLC25A3', 'RBM42', 'RPRD1B', 'CCT7', 'DENR', 'CYCS', 'ACTR1B', 'NLE1', 'ACTB', 'MRPL20', 'DCUN1D5', 'WARS', 'RTCB', 'TP53I13', 'BRF1', 'NAA25', 'RPL23', 'SUPT20H', 'UBA5', 'MCM3', 'NUP62', 'TSEN2', 'NDUFV2', 'UQCRH', 'SERBP1', 'EIF3A', 'ZMAT5', 'ATP6V1C1', 'AIFM1', 'PDCD11', 'TMED2', 'NSUN4', 'DUT', 'CRCP', 'MRPS23', 'CMPK1', 'SNAPC1', 'MIS18BP1', 'SPTLC1', 'RNF168', 'ZW10', 'ETF1', 'RPL26L1', 'IPO13', 'ARL2', 'RPL10A', 'DPH2', 'MRPL1', 'CENPN', 'TOP3A', 'DSN1', 'MCM4', 'VPS37A', 'TUFM', 'RCL1', 'STIL', 'RPL24', 'RBM25', 'GBF1', 'DNAJC17', 'NISCH', 'CLOCK', 'TUBA1B', 'GTF2H2', 'GRPEL1', 'PPIL1', 'SMG7', 'UBQLN4', 'RBM4', 'DBF4', 'PPP2CA', 'MIS18A', 'YKT6', 'SDHC', 'TWF1', 'TOX4', 'RPS27A', 'STRIP1', 'AURKA', 'DYNC1I2', 'SUPV3L1', 'EGLN2', 'OIP5', 'NAA20', 'CINP', 'POLD3', 'OSTC', 'ITGB1BP1', 'PPP2CB', 'BUB1', 'ZNHIT1', 'DIMT1', 'HEXIM1', 'TSFM', 'POLR2M', 'RPL35', 'RAD51D', 'HSPE1', 'COA5', 'CPNE7', 'FDXR', 'STXBP4', 'GINS4', 'PAM16', 'MCM2', 'FKBP9', 'GYG1', 'TTK', 'RPS27', 'ANKS6', 'PPIL2', 'TTC4', 'MRPL14', 'TMX2', 'LAMTOR1', 'RUVBL2', 'ALG14', 'KEAP1', 'CPOX', 'TOMM20', 'CCDC137', 'MRPS21', 'TUBA1C', 'TRMT5', 'CDC73', 'BUD13', 'DNAJA1', 'SMARCE1', 'DMAP1', 'SNUPN', 'NAPG', 'CENPE', 'INO80B', 'RTRAF', 'SUPT16H', 'CPSF3', 'CUL1', 'CUL3', 'CCNH', 'EIF4H', 'PPME1', 'ZNF787', 'RAC3', 'MSL1', 'ATP6V1B2', 'SPG7', 'DPY19L4', 'SPC24', 'OPA1', 'CABIN1', 'POLL', 'SNRPB', 'PRPF39', 'NSMCE1', 'RPS19', 'SLC39A10', 'TMEM199', 'YARS', 'EIF4E', 'CCT4', 'DYNLL1', 'SAE1', 'ANAPC1', 'MRPS34', 'HNRNPU', 'COX7C', 'NUP85', 'METTL1', 'UROD', 'SUMO2', 'WDR44', 'IMPA2', 'IST1', 'MRPL32', 'PEF1', 'RNF14', 'TTI1', 'EIF2S2', 'PPIL4', 'ZBTB17', 'RNPC3', 'TBCB', 'GPN2', 'VPS41', 'SRPRB', 'COQ5', 'POLD1', 'WDR33', 'UMPS', 'TUBG1', 'METTL23', 'CCP110', 'MRPL49', 'CDC6', 'ARL4D', 'CS', 'RPA1', 'HMGCS1', 'NUB1', 'GMPPB', 'SNRNP27', 'NCAPD2', 'CHCHD2', 'NFYB', 'RNF20', 'EIF1', 'PTK2', 'ZNF236', 'VPS28', 'TFDP1', 'C9orf16', 'PDCD6', 'GART', 'PSME1', 'YEATS4', 'U2SURP', 'TNPO3', 'CENPC', 'ERCC2', 'ATP5MF', 'POLD2', 'CLTC', 'EEF1G', 'ISCA2', 'RNMT', 'DNAJC19', 'NSF', 'CHMP7', 'DHX33', 'SARS2', 'DSTYK', 'DPAGT1', 'UBE2M', 'MRPL22', 'CENPT', 'NDUFA8', 'INTS13', 'UQCRC2', 'GPKOW', 'ATF5', 'BRIX1', 'SASS6', 'ACTR2', 'HAUS4', 'CTU2', 'POLRMT', 'MRPS7', 'MMS22L', 'EIF3L', 'WDR4', 'TMEM127', 'ELP6', 'RPL15', 'GSK3B', 'MYBL2', 'WBP1', 'MRPL37', 'GTF3C6', 'AP2M1', 'DDX17', 'SET', 'EIF3M', 'PRORP', 'RNGTT', 'MFAP1', 'HJURP', 'MRPS14', 'EPB41L2', 'WDR5', 'BUB1B', 'RPL32', 'TRIAP1', 'FASTKD5', 'CCDC84', 'COG1', 'FDPS', 'DNAJC8', 'FUNDC2', 'EXOC7', 'NUDCD3', 'XRN1', 'MBIP', 'YARS2', 'DDX55', 'URI1', 'NDUFB10', 'EIF3I', 'TUBGCP4', 'SLC2A8', 'TTC1', 'MPHOSPH6', 'UFM1', 'UBE2I', 'CBLL1', 'PITRM1', 'NUP153', 'VPS33A', 'MRPS26', 'EIF2B3', 'CYC1', 'CEP97', 'SEPHS2', 'PLEKHN1', 'EXOSC1', 'INTS14', 'MCM5', 'AGBL5', 'WDR83OS', 'ARMC6', 'AFG3L2', 'RHOQ', 'HBS1L', 'RPL35A', 'PDCD5', 'AASDHPPT', 'FAM136A', 'HPS5', 'HDAC3', 'YPEL5', 'RPS2', 'SEC63', 'NDUFA11', 'SRRT', 'ADAM10', 'GET3', 'TUBE1', 'EIF6', 'NUP214', 'NDUFB8', 'RPL5', 'PET117', 'NAA50', 'SPCS2', 'SNAPC3', 'ELL', 'PRIM2', 'MRPL34', 'TLCD1', 'CCT6A', 'TIMM10', 'RPL26', 'MVD', 'HMGB3', 'CIAO1', 'ADRM1', 'HNRNPA1', 'SNRNP25', 'FXN', 'MTOR', 'IWS1', 'SSBP3', 'BDP1', 'COMTD1', 'MICOS10', 'RPL39', 'HSD17B10', 'SMG8', 'DPH6', 'DCTN4', 'PTEN', 'PHB2', 'MED11', 'RAB18', 'PES1', 'NKAPD1', 'NUDC', 'RPS3', 'CDK1', 'COG6', 'EXOSC4', 'LAS1L', 'RPP14', 'EXOSC8', 'RPS6', 'NOM1', 'DDX1', 'FBXW7', 'KCMF1', 'PTPN1', 'TXN', 'RPL9', 'MRPL19', 'MRPL28', 'GNPNAT1', 'PPA2', 'PGK1', 'SNRPG', 'SAMD4B', 'NOLC1', 'EEFSEC', 'AHCTF1', 'CLNS1A', 'TAF8', 'DOHH', 'NUP35', 'RPUSD4', 'EP400', 'IFITM2', 'THOC7']
+# train gets all other perturbations automatically

data/state_replogle_filtered_analysis/rpe1_zeroshot.toml

@@ -0,0 +1,14 @@
+[datasets]
+replogle = "/data/replogle_nogwps_v2" # ADDS ALL h5 or h5ad files in this folder to training
+
+# Training specifications
+# All cell types in a dataset automatically go into training (excluding zeroshot/fewshot overrides)
+[training]
+replogle = "train"
+
+# Zeroshot specifications - entire cell types go to val or test
+[zeroshot]
+"replogle.rpe1" = "test"
+
+# Fewshot specifications - explicit perturbation lists
+[fewshot]

data/state_replogle_filtered_reembedded/replogle_concat.h5ad

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac9644b41e9bd84ff00b029184c210f03363eb31e193f1c1713f75b7b3ba6ee3
+size 30010775846