Remove W&B tracking and personal paths; complete BioMod naming
Browse filesRemove W&B logging, account settings, CLI options and direct dependencies while preserving local logs, checkpoint saving and resume behavior. Replace historical workstation paths in two dataset summaries with repository-relative paths and refresh their manifest checksum rows. Include the previously reviewed BioMod naming changes required by the updated pretraining entrypoint; retain legacy checkpoint compatibility. Model weights, paired model configuration, original dataset payloads and unrelated files are unchanged.
- BioMod/biomod.py +12 -132
- BioMod/cell_annotation/train.py +4 -3
- BioMod/command.txt +0 -1
- BioMod/environment.yml +0 -1
- BioMod/gene_perturbation/components/model.py +3 -3
- BioMod/gene_perturbation/components/trainer.py +1 -29
- BioMod/gene_perturbation/train.py +0 -2
- BioMod/gene_perturbation/tutorial.ipynb +0 -1
- BioMod/models_biomod.py +0 -0
- BioMod/requirements.txt +0 -1
- data/downstream/final_data/drug_combination_synergy/MANIFEST.csv +1 -1
- data/downstream/final_data/drug_combination_synergy/SUMMARY.json +1 -1
- data/downstream/final_data/single_drug_response/MANIFEST.csv +1 -1
- data/downstream/final_data/single_drug_response/SUMMARY.json +1 -1
BioMod/biomod.py
CHANGED
|
@@ -65,6 +65,7 @@
|
|
| 65 |
sys.path.insert(0, str(SOURCE_DIR))
|
| 66 |
from models_biomod import (
|
| 67 |
BioMod,
|
|
|
|
| 68 |
sanitize_biomod_config,
|
| 69 |
)
|
| 70 |
from biomod_dataset import BioModSCDataset, load_biomod_data
|
|
@@ -81,8 +82,6 @@
|
|
| 81 |
from torch import autocast
|
| 82 |
from torch.cuda.amp import GradScaler
|
| 83 |
|
| 84 |
-
import wandb
|
| 85 |
-
|
| 86 |
# ========== [Step 0: Parameter Analysis] ==========
|
| 87 |
# Example: Assume the default parameters batch_size=3, gene_num=27874, mask_ratio=0.75, grad_acc=60
|
| 88 |
parser = argparse.ArgumentParser()
|
|
@@ -116,14 +115,14 @@
|
|
| 116 |
help='Reconstruct loss aggregation method: gene_mean=masked gene equal weight (current); module_mean=masked module equal weight (MAE style)')
|
| 117 |
parser.add_argument("--recon_target_level", type=str, default="gene", choices=["gene", "module_mean_value"],
|
| 118 |
help='Reconstruction target granularity: gene=gene-by-gene reconstruction; module_mean_value=decoder directly outputs the module mean and calculates the loss with the target mean')
|
| 119 |
-
parser.add_argument("--decoder_type", type=
|
| 120 |
-
help='decoder type: none=directly use encoder to output predictions; simple=no token mixing;
|
| 121 |
parser.add_argument("--decoder_gene_agg", action='store_true',
|
| 122 |
help='The reconstruction head is changed to module->gene aggregation and then gene-level reconstruction (off by default to facilitate comparison with the old module_head)')
|
| 123 |
parser.add_argument("--dec_embed_dim", type=int, default=512, help='decoder token dimension (can be smaller than encoder base_dim, default 512)')
|
| 124 |
-
parser.add_argument("--dec_depth", type=int, default=4, help='Decoder layer number (valid when
|
| 125 |
-
parser.add_argument("--dec_heads", type=int, default=8, help='Number of decoder attention heads (valid when
|
| 126 |
-
parser.add_argument("--dec_dim_head", type=int, default=64, help='
|
| 127 |
parser.add_argument("--ckpt_dir", type=str, default=str(PROJECT_ROOT / 'tutorial_runs' / 'pretraining'), help='Directory for generated checkpoints; user relative paths use the current working directory.')
|
| 128 |
parser.add_argument("--model_name", type=str, default='biomod_27k', help='Pretrained model name.')
|
| 129 |
parser.add_argument("--data_path", type=str, required=True,
|
|
@@ -140,7 +139,6 @@
|
|
| 140 |
parser.add_argument("--fusion_method", type=str, default="multiply",
|
| 141 |
choices=["multiply", "add", "gated_residual"],
|
| 142 |
help='Fusion method: "multiply"=multiply; "add"=add to gene2vec after expressing MLP; "gated_residual"=inject expression information into the gated residual')
|
| 143 |
-
parser.add_argument("-use_wandb", action='store_true', default=False)
|
| 144 |
parser.add_argument("-compile", action='store_true')
|
| 145 |
parser.add_argument("--model_path", type=str, default='None',
|
| 146 |
help='Checkpoint file to load; use its matching configuration for continued pretraining.')
|
|
@@ -160,10 +158,6 @@
|
|
| 160 |
parser.add_argument("--latest_ckpt_every", type=int, default=0,
|
| 161 |
help='The latest checkpoint saves the cadence; 0 means closed (only timing and epoch are saved)')
|
| 162 |
parser.add_argument("--amp_dtype", type=str, default="bfloat16")
|
| 163 |
-
parser.add_argument("--api_key", type=str, default="None")
|
| 164 |
-
parser.add_argument("--wandb_resume", action='store_true', help='Resume wandb run from previous checkpoint')
|
| 165 |
-
parser.add_argument("--wandb_run_id", type=str, default=None, help='WandB run ID to resume (e.g., nbi8qzmn)')
|
| 166 |
-
parser.add_argument("--resume_train_log_step", type=int, default=-1, help='Resume train_log_step from this value (set to override auto-calculation)')# Not in use yet
|
| 167 |
parser.add_argument("--load_opt_and_sched", action='store_true')
|
| 168 |
parser.add_argument("--lr_step_within_epoch", action='store_true',
|
| 169 |
help='Update the learning rate according to the optimizer update step; the cycle length is updates_per_epoch (gradually decreasing in each epoch)')
|
|
@@ -311,82 +305,13 @@ def restore_model_args_from_config(args, config):
|
|
| 311 |
elif args.pretrain_arch == "encdec":
|
| 312 |
if args.mask_style != "replace":
|
| 313 |
raise ValueError("pretrain_arch=encdec requires --mask_style replace")
|
| 314 |
-
if args.decoder_type not in {"
|
| 315 |
-
raise ValueError("pretrain_arch=encdec requires decoder_type in {'
|
| 316 |
|
| 317 |
local_rank = args.local_rank
|
| 318 |
rank = int(os.environ["RANK"])
|
| 319 |
is_master = rank == 0
|
| 320 |
|
| 321 |
-
if is_master and args.use_wandb:
|
| 322 |
-
api_key = args.api_key
|
| 323 |
-
if api_key != "None":
|
| 324 |
-
wandb.login(key=api_key)
|
| 325 |
-
|
| 326 |
-
# WandB resume logic
|
| 327 |
-
wandb_init_kwargs = {
|
| 328 |
-
"project": "pretrain_biomod",
|
| 329 |
-
"name": args.model_name,
|
| 330 |
-
"entity": "icml_2023_mlb",
|
| 331 |
-
"config": {
|
| 332 |
-
"architecture": "biomod",
|
| 333 |
-
"gene_num": args.gene_num,
|
| 334 |
-
"cell_num": args.cell_num,
|
| 335 |
-
"load model": args.model_path,
|
| 336 |
-
"start epoch": args.start_epoch,
|
| 337 |
-
"epochs to train": args.epoch,
|
| 338 |
-
"valid_every": args.valid_every,
|
| 339 |
-
"batch_size": args.batch_size,
|
| 340 |
-
"mask_ratio": args.mask_ratio,
|
| 341 |
-
"pretrain_arch": args.pretrain_arch,
|
| 342 |
-
"mask_style": args.mask_style,
|
| 343 |
-
"mask_token_mode": args.mask_token_mode,
|
| 344 |
-
"pos_embed_mode": args.pos_embed_mode,
|
| 345 |
-
"loss_mode": args.loss_mode,
|
| 346 |
-
"recon_target_level": args.recon_target_level,
|
| 347 |
-
"decoder_type": args.decoder_type,
|
| 348 |
-
"decoder_gene_agg": args.decoder_gene_agg,
|
| 349 |
-
"dec_embed_dim": args.dec_embed_dim,
|
| 350 |
-
"use_cls_token": args.use_cls_token,
|
| 351 |
-
"cls_loss_weight": args.cls_loss_weight if args.use_cls_token else 0.0,
|
| 352 |
-
"cls_head_hidden_dim": args.cls_head_hidden_dim,
|
| 353 |
-
"use_module": True,
|
| 354 |
-
"use_go": False,
|
| 355 |
-
"encoder_only": True,
|
| 356 |
-
"shuffle_module_genes": args.shuffle_module_genes,
|
| 357 |
-
"save_init_ckpt_and_exit": args.save_init_ckpt_and_exit,
|
| 358 |
-
}
|
| 359 |
-
}
|
| 360 |
-
|
| 361 |
-
# Support resuming previous wandb run
|
| 362 |
-
if args.wandb_resume:
|
| 363 |
-
if args.wandb_run_id:
|
| 364 |
-
wandb_init_kwargs["id"] = args.wandb_run_id
|
| 365 |
-
wandb_init_kwargs["resume"] = "must"
|
| 366 |
-
print(f"[WandB] Resuming run with ID: {args.wandb_run_id}")
|
| 367 |
-
else:
|
| 368 |
-
wandb_init_kwargs["resume"] = "allow"
|
| 369 |
-
print("[WandB] Resuming run (auto-detect)")
|
| 370 |
-
|
| 371 |
-
wandb.init(**wandb_init_kwargs)
|
| 372 |
-
# ========== wandb indicator definition ==========
|
| 373 |
-
# Define a custom x-axis so that epoch-level metrics have epoch as the abscissa
|
| 374 |
-
wandb.define_metric("epoch")
|
| 375 |
-
# Epoch level loss (put on the same picture for easy comparison)
|
| 376 |
-
wandb.define_metric("train_epoch_loss", step_metric="epoch")
|
| 377 |
-
wandb.define_metric("val_epoch_loss", step_metric="epoch")
|
| 378 |
-
wandb.define_metric("val_in_train_loss", step_metric="epoch")
|
| 379 |
-
# step level metrics (training process)
|
| 380 |
-
wandb.define_metric("train_log_step")
|
| 381 |
-
wandb.define_metric("train_loss", step_metric="train_log_step")
|
| 382 |
-
wandb.define_metric("learning_rate", step_metric="train_log_step")
|
| 383 |
-
|
| 384 |
-
# ===== Old code (commented, valid_log_step sharing issue exists) =====
|
| 385 |
-
# wandb.define_metric("valid_log_step")
|
| 386 |
-
# wandb.define_metric("val_step_loss", step_metric="valid_log_step")
|
| 387 |
-
# wandb.define_metric("val_loss", step_metric="valid_log_step")
|
| 388 |
-
# ========================================================
|
| 389 |
-
|
| 390 |
SEED = args.seed
|
| 391 |
EPOCHS = args.epoch
|
| 392 |
BATCH_SIZE = args.batch_size # Example value: 3
|
|
@@ -576,7 +501,7 @@ def resolve_total_cells_for_split():
|
|
| 576 |
"use_gene2vec": use_gene2vec, # Whether to use pretrained gene2vec (ablation experiment)
|
| 577 |
"loss_mode": args.loss_mode, # Loss aggregation method: gene_mean / module_mean
|
| 578 |
"recon_target_level": args.recon_target_level, # Reconstruction target granularity: gene/module_mean_value
|
| 579 |
-
"decoder_type": args.decoder_type, # decoder type: simple /
|
| 580 |
"decoder_gene_agg": args.decoder_gene_agg, # True: module->gene aggregation and reconstruction; False: module_head reconstruction
|
| 581 |
"dec_embed_dim": args.dec_embed_dim,
|
| 582 |
"dec_depth": args.dec_depth,
|
|
@@ -690,8 +615,6 @@ def save_initialized_checkpoint_and_exit(model_obj):
|
|
| 690 |
f"contains loaded weights rather than random initialization."
|
| 691 |
)
|
| 692 |
dist.barrier()
|
| 693 |
-
if is_master and args.use_wandb:
|
| 694 |
-
wandb.finish()
|
| 695 |
dist.destroy_process_group()
|
| 696 |
raise SystemExit(0)
|
| 697 |
|
|
@@ -825,22 +748,6 @@ def any_rank_true(local_flag: bool) -> bool:
|
|
| 825 |
model = torch.compile(model)
|
| 826 |
|
| 827 |
# ========== [Step 4: Training Cycle] ==========
|
| 828 |
-
# train_log_step: training step counter, used for the x-axis of wandb logs
|
| 829 |
-
if args.resume_train_log_step >= 0:
|
| 830 |
-
# User manually specifies the connection location
|
| 831 |
-
train_log_step = args.resume_train_log_step
|
| 832 |
-
if is_master:
|
| 833 |
-
print(f"[Resume] Using specified train_log_step: {train_log_step}")
|
| 834 |
-
elif args.start_epoch > 0:
|
| 835 |
-
# Automatic calculation: Each epoch has len(train_loader) steps, recorded once every LOG_EVERY step
|
| 836 |
-
steps_per_epoch = len(train_loader)
|
| 837 |
-
train_log_step = (args.start_epoch * steps_per_epoch) // LOG_EVERY
|
| 838 |
-
if is_master:
|
| 839 |
-
print(f"[Resume] Calculated train_log_step: {train_log_step} from start_epoch: {args.start_epoch}")
|
| 840 |
-
else:
|
| 841 |
-
train_log_step = 0
|
| 842 |
-
# valid_log_step = 0 #Old code (annotated): shared by step verification and epoch verification, resulting in confusing curves
|
| 843 |
-
|
| 844 |
assert EPOCHS != 0
|
| 845 |
dist.barrier()
|
| 846 |
|
|
@@ -955,18 +862,7 @@ def any_rank_true(local_flag: bool) -> bool:
|
|
| 955 |
)
|
| 956 |
else:
|
| 957 |
print(f"- epoch {i} step {index} train_loss {loss.item()} -")
|
| 958 |
-
if is_master
|
| 959 |
-
wandb_payload = {
|
| 960 |
-
"train_loss": loss.item(),
|
| 961 |
-
"learning_rate": scheduler.optimizer.param_groups[0]['lr'],
|
| 962 |
-
"train_log_step": train_log_step,
|
| 963 |
-
}
|
| 964 |
-
loss_breakdown = get_loss_breakdown(model)
|
| 965 |
-
if loss_breakdown:
|
| 966 |
-
wandb_payload["train_recon_loss"] = loss_breakdown.get("recon_loss", 0.0)
|
| 967 |
-
wandb_payload["train_cls_loss"] = loss_breakdown.get("cls_loss", 0.0)
|
| 968 |
-
wandb.log(wandb_payload)# train every LOG_EVERY step train_log_step train_loss, learning_rate
|
| 969 |
-
train_log_step += 1
|
| 970 |
if args.latest_ckpt_every > 0 and (index % args.latest_ckpt_every == 0):
|
| 971 |
latest_ckpt_dir = os.path.join(ckpt_dir, 'latest')
|
| 972 |
os.makedirs(latest_ckpt_dir, exist_ok=True)
|
|
@@ -1005,10 +901,6 @@ def any_rank_true(local_flag: bool) -> bool:
|
|
| 1005 |
|
| 1006 |
# Step 5.2: Print verification progress regularly
|
| 1007 |
if val_index % LOG_EVERY == 0:
|
| 1008 |
-
# Commented out old code: each step of verification loss is recorded to wandb (leading to confusing curves)
|
| 1009 |
-
# if is_master and args.use_wandb:
|
| 1010 |
-
# wandb.log({"val_step_loss": loss_val.item(), "valid_log_step": valid_log_step})
|
| 1011 |
-
# valid_log_step += 1
|
| 1012 |
if is_master:
|
| 1013 |
current_time = time.time()
|
| 1014 |
if current_time - last_save_time >= save_time_interval:
|
|
@@ -1058,14 +950,11 @@ def any_rank_true(local_flag: bool) -> bool:
|
|
| 1058 |
# Make sure all processes are in sync
|
| 1059 |
dist.barrier()
|
| 1060 |
|
| 1061 |
-
# Step 5.5: Print
|
| 1062 |
if is_master:
|
| 1063 |
current_time = time.time()
|
| 1064 |
time_stamp = datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S')
|
| 1065 |
print(f' == Epoch: {i} Step {index} Time {time_stamp} | Validation Loss: {loss_val_reduced:.6f}')
|
| 1066 |
-
if args.use_wandb:
|
| 1067 |
-
# val_in_train_loss: The average loss verified during training, with epoch as the x-axis
|
| 1068 |
-
wandb.log({"val_in_train_loss": loss_val_reduced, "epoch": i})
|
| 1069 |
model.train() # Switch back to training mode
|
| 1070 |
dist.barrier()
|
| 1071 |
|
|
@@ -1075,8 +964,6 @@ def any_rank_true(local_flag: bool) -> bool:
|
|
| 1075 |
epoch_loss = get_reduced(epoch_loss, local_rank, 0, world_size) # Multi-process average
|
| 1076 |
if is_master:
|
| 1077 |
print(f' == Epoch: {i} | Training Loss: {epoch_loss:.6f}')
|
| 1078 |
-
if args.use_wandb:
|
| 1079 |
-
wandb.log({"train_epoch_loss": epoch_loss, "epoch": i}) # Epoch training loss, with epoch as the x-axis
|
| 1080 |
dist.barrier()
|
| 1081 |
if not args.lr_step_within_epoch:
|
| 1082 |
scheduler.step() # Update learning rate
|
|
@@ -1114,10 +1001,6 @@ def any_rank_true(local_flag: bool) -> bool:
|
|
| 1114 |
if index % LOG_EVERY == 0:
|
| 1115 |
if is_master:
|
| 1116 |
print(f"- epoch {i} step {index} val_loss {loss.item()} -")
|
| 1117 |
-
# Commented out old code: each step of verification loss is recorded to wandb (leading to confusing curves)
|
| 1118 |
-
# if is_master and args.use_wandb:
|
| 1119 |
-
# wandb.log({"val_loss": loss.item(), "valid_log_step": valid_log_step})
|
| 1120 |
-
# valid_log_step += 1
|
| 1121 |
if is_master:
|
| 1122 |
current_time = time.time()
|
| 1123 |
if current_time - last_save_time >= save_time_interval:
|
|
@@ -1162,13 +1045,10 @@ def any_rank_true(local_flag: bool) -> bool:
|
|
| 1162 |
finally:
|
| 1163 |
dist.barrier()
|
| 1164 |
|
| 1165 |
-
# Step 7.6: Print epoch
|
| 1166 |
if is_master:
|
| 1167 |
t_epoch_end = time.time()
|
| 1168 |
print(f' == Epoch: {i} | Validation Loss: {val_loss:.6f} | epoch_cost_time: {t_epoch_end-t_epoch_start:.3f}')
|
| 1169 |
-
if args.use_wandb:
|
| 1170 |
-
# val_epoch_loss: The average loss of epoch verification, with epoch as the x-axis
|
| 1171 |
-
wandb.log({"val_epoch_loss": val_loss, "epoch": i})
|
| 1172 |
|
| 1173 |
# ========== [Step 8: Save Checkpoint] ==========
|
| 1174 |
# Save model checkpoint at the end of each epoch
|
|
|
|
| 65 |
sys.path.insert(0, str(SOURCE_DIR))
|
| 66 |
from models_biomod import (
|
| 67 |
BioMod,
|
| 68 |
+
normalize_biomod_decoder_type,
|
| 69 |
sanitize_biomod_config,
|
| 70 |
)
|
| 71 |
from biomod_dataset import BioModSCDataset, load_biomod_data
|
|
|
|
| 82 |
from torch import autocast
|
| 83 |
from torch.cuda.amp import GradScaler
|
| 84 |
|
|
|
|
|
|
|
| 85 |
# ========== [Step 0: Parameter Analysis] ==========
|
| 86 |
# Example: Assume the default parameters batch_size=3, gene_num=27874, mask_ratio=0.75, grad_acc=60
|
| 87 |
parser = argparse.ArgumentParser()
|
|
|
|
| 115 |
help='Reconstruct loss aggregation method: gene_mean=masked gene equal weight (current); module_mean=masked module equal weight (MAE style)')
|
| 116 |
parser.add_argument("--recon_target_level", type=str, default="gene", choices=["gene", "module_mean_value"],
|
| 117 |
help='Reconstruction target granularity: gene=gene-by-gene reconstruction; module_mean_value=decoder directly outputs the module mean and calculates the loss with the target mean')
|
| 118 |
+
parser.add_argument("--decoder_type", type=normalize_biomod_decoder_type, default="biomod", choices=["none", "simple", "biomod", "cross_attn"],
|
| 119 |
+
help='decoder type: none=directly use encoder to output predictions; simple=no token mixing; biomod=decoder self-attention; cross_attn=only masked->visible cross-attention')
|
| 120 |
parser.add_argument("--decoder_gene_agg", action='store_true',
|
| 121 |
help='The reconstruction head is changed to module->gene aggregation and then gene-level reconstruction (off by default to facilitate comparison with the old module_head)')
|
| 122 |
parser.add_argument("--dec_embed_dim", type=int, default=512, help='decoder token dimension (can be smaller than encoder base_dim, default 512)')
|
| 123 |
+
parser.add_argument("--dec_depth", type=int, default=4, help='Decoder layer number (valid when biomod/cross_attn)')
|
| 124 |
+
parser.add_argument("--dec_heads", type=int, default=8, help='Number of decoder attention heads (valid when biomod/cross_attn)')
|
| 125 |
+
parser.add_argument("--dec_dim_head", type=int, default=64, help='BioMod decoder dimension per head (ignored when cross_attn)')
|
| 126 |
parser.add_argument("--ckpt_dir", type=str, default=str(PROJECT_ROOT / 'tutorial_runs' / 'pretraining'), help='Directory for generated checkpoints; user relative paths use the current working directory.')
|
| 127 |
parser.add_argument("--model_name", type=str, default='biomod_27k', help='Pretrained model name.')
|
| 128 |
parser.add_argument("--data_path", type=str, required=True,
|
|
|
|
| 139 |
parser.add_argument("--fusion_method", type=str, default="multiply",
|
| 140 |
choices=["multiply", "add", "gated_residual"],
|
| 141 |
help='Fusion method: "multiply"=multiply; "add"=add to gene2vec after expressing MLP; "gated_residual"=inject expression information into the gated residual')
|
|
|
|
| 142 |
parser.add_argument("-compile", action='store_true')
|
| 143 |
parser.add_argument("--model_path", type=str, default='None',
|
| 144 |
help='Checkpoint file to load; use its matching configuration for continued pretraining.')
|
|
|
|
| 158 |
parser.add_argument("--latest_ckpt_every", type=int, default=0,
|
| 159 |
help='The latest checkpoint saves the cadence; 0 means closed (only timing and epoch are saved)')
|
| 160 |
parser.add_argument("--amp_dtype", type=str, default="bfloat16")
|
|
|
|
|
|
|
|
|
|
|
|
|
| 161 |
parser.add_argument("--load_opt_and_sched", action='store_true')
|
| 162 |
parser.add_argument("--lr_step_within_epoch", action='store_true',
|
| 163 |
help='Update the learning rate according to the optimizer update step; the cycle length is updates_per_epoch (gradually decreasing in each epoch)')
|
|
|
|
| 305 |
elif args.pretrain_arch == "encdec":
|
| 306 |
if args.mask_style != "replace":
|
| 307 |
raise ValueError("pretrain_arch=encdec requires --mask_style replace")
|
| 308 |
+
if args.decoder_type not in {"biomod", "cross_attn"}:
|
| 309 |
+
raise ValueError("pretrain_arch=encdec requires decoder_type in {'biomod', 'cross_attn'}")
|
| 310 |
|
| 311 |
local_rank = args.local_rank
|
| 312 |
rank = int(os.environ["RANK"])
|
| 313 |
is_master = rank == 0
|
| 314 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 315 |
SEED = args.seed
|
| 316 |
EPOCHS = args.epoch
|
| 317 |
BATCH_SIZE = args.batch_size # Example value: 3
|
|
|
|
| 501 |
"use_gene2vec": use_gene2vec, # Whether to use pretrained gene2vec (ablation experiment)
|
| 502 |
"loss_mode": args.loss_mode, # Loss aggregation method: gene_mean / module_mean
|
| 503 |
"recon_target_level": args.recon_target_level, # Reconstruction target granularity: gene/module_mean_value
|
| 504 |
+
"decoder_type": args.decoder_type, # decoder type: simple / biomod
|
| 505 |
"decoder_gene_agg": args.decoder_gene_agg, # True: module->gene aggregation and reconstruction; False: module_head reconstruction
|
| 506 |
"dec_embed_dim": args.dec_embed_dim,
|
| 507 |
"dec_depth": args.dec_depth,
|
|
|
|
| 615 |
f"contains loaded weights rather than random initialization."
|
| 616 |
)
|
| 617 |
dist.barrier()
|
|
|
|
|
|
|
| 618 |
dist.destroy_process_group()
|
| 619 |
raise SystemExit(0)
|
| 620 |
|
|
|
|
| 748 |
model = torch.compile(model)
|
| 749 |
|
| 750 |
# ========== [Step 4: Training Cycle] ==========
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 751 |
assert EPOCHS != 0
|
| 752 |
dist.barrier()
|
| 753 |
|
|
|
|
| 862 |
)
|
| 863 |
else:
|
| 864 |
print(f"- epoch {i} step {index} train_loss {loss.item()} -")
|
| 865 |
+
if is_master:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 866 |
if args.latest_ckpt_every > 0 and (index % args.latest_ckpt_every == 0):
|
| 867 |
latest_ckpt_dir = os.path.join(ckpt_dir, 'latest')
|
| 868 |
os.makedirs(latest_ckpt_dir, exist_ok=True)
|
|
|
|
| 901 |
|
| 902 |
# Step 5.2: Print verification progress regularly
|
| 903 |
if val_index % LOG_EVERY == 0:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 904 |
if is_master:
|
| 905 |
current_time = time.time()
|
| 906 |
if current_time - last_save_time >= save_time_interval:
|
|
|
|
| 950 |
# Make sure all processes are in sync
|
| 951 |
dist.barrier()
|
| 952 |
|
| 953 |
+
# Step 5.5: Print validation results
|
| 954 |
if is_master:
|
| 955 |
current_time = time.time()
|
| 956 |
time_stamp = datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S')
|
| 957 |
print(f' == Epoch: {i} Step {index} Time {time_stamp} | Validation Loss: {loss_val_reduced:.6f}')
|
|
|
|
|
|
|
|
|
|
| 958 |
model.train() # Switch back to training mode
|
| 959 |
dist.barrier()
|
| 960 |
|
|
|
|
| 964 |
epoch_loss = get_reduced(epoch_loss, local_rank, 0, world_size) # Multi-process average
|
| 965 |
if is_master:
|
| 966 |
print(f' == Epoch: {i} | Training Loss: {epoch_loss:.6f}')
|
|
|
|
|
|
|
| 967 |
dist.barrier()
|
| 968 |
if not args.lr_step_within_epoch:
|
| 969 |
scheduler.step() # Update learning rate
|
|
|
|
| 1001 |
if index % LOG_EVERY == 0:
|
| 1002 |
if is_master:
|
| 1003 |
print(f"- epoch {i} step {index} val_loss {loss.item()} -")
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1004 |
if is_master:
|
| 1005 |
current_time = time.time()
|
| 1006 |
if current_time - last_save_time >= save_time_interval:
|
|
|
|
| 1045 |
finally:
|
| 1046 |
dist.barrier()
|
| 1047 |
|
| 1048 |
+
# Step 7.6: Print epoch validation results
|
| 1049 |
if is_master:
|
| 1050 |
t_epoch_end = time.time()
|
| 1051 |
print(f' == Epoch: {i} | Validation Loss: {val_loss:.6f} | epoch_cost_time: {t_epoch_end-t_epoch_start:.3f}')
|
|
|
|
|
|
|
|
|
|
| 1052 |
|
| 1053 |
# ========== [Step 8: Save Checkpoint] ==========
|
| 1054 |
# Save model checkpoint at the end of each epoch
|
BioMod/cell_annotation/train.py
CHANGED
|
@@ -695,7 +695,7 @@ def infer_model_config(args, checkpoint: dict, pkl_config: Optional[dict], biomo
|
|
| 695 |
elif isinstance(checkpoint, dict) and "config" in checkpoint:
|
| 696 |
config = sanitize_biomod_config(
|
| 697 |
checkpoint["config"],
|
| 698 |
-
source=str(resolve_path(args.
|
| 699 |
log_fn=logger.info,
|
| 700 |
)
|
| 701 |
logger.info("Using config from checkpoint")
|
|
@@ -767,7 +767,7 @@ def load_backbone(args, device: torch.device) -> Tuple["BioMod", List[List[int]]
|
|
| 767 |
else:
|
| 768 |
logger.warning("Config PKL not found at %s; fallback to checkpoint config", config_path)
|
| 769 |
|
| 770 |
-
ckpt_path = resolve_path(args.
|
| 771 |
checkpoint = torch.load(ckpt_path, map_location=device)
|
| 772 |
|
| 773 |
biomod_data = None
|
|
@@ -1173,11 +1173,12 @@ def parse_args():
|
|
| 1173 |
parser.add_argument("--max_cells", type=int, default=None)
|
| 1174 |
|
| 1175 |
parser.add_argument("--config_pkl", type=str, default=str(WEIGHTS_DIR / "BioMod.pkl"))
|
|
|
|
| 1176 |
parser.add_argument(
|
| 1177 |
"--biomod_checkpoint", "--performer_checkpoint",
|
| 1178 |
type=str,
|
| 1179 |
default=str(WEIGHTS_DIR / "BioMod.pth"),
|
| 1180 |
-
dest='
|
| 1181 |
)
|
| 1182 |
parser.add_argument("--biomod_data_dir", "--module_data_dir", type=str, default=str(ASSET_DIR / "biomod_data_all"), dest='module_data_dir')
|
| 1183 |
parser.add_argument("--gene_list_path", type=str, default=str(ASSET_DIR / "selected_genes_27k.txt"))
|
|
|
|
| 695 |
elif isinstance(checkpoint, dict) and "config" in checkpoint:
|
| 696 |
config = sanitize_biomod_config(
|
| 697 |
checkpoint["config"],
|
| 698 |
+
source=str(resolve_path(args.biomod_checkpoint)),
|
| 699 |
log_fn=logger.info,
|
| 700 |
)
|
| 701 |
logger.info("Using config from checkpoint")
|
|
|
|
| 767 |
else:
|
| 768 |
logger.warning("Config PKL not found at %s; fallback to checkpoint config", config_path)
|
| 769 |
|
| 770 |
+
ckpt_path = resolve_path(args.biomod_checkpoint)
|
| 771 |
checkpoint = torch.load(ckpt_path, map_location=device)
|
| 772 |
|
| 773 |
biomod_data = None
|
|
|
|
| 1173 |
parser.add_argument("--max_cells", type=int, default=None)
|
| 1174 |
|
| 1175 |
parser.add_argument("--config_pkl", type=str, default=str(WEIGHTS_DIR / "BioMod.pkl"))
|
| 1176 |
+
# Retain the legacy CLI spelling as a compatibility alias.
|
| 1177 |
parser.add_argument(
|
| 1178 |
"--biomod_checkpoint", "--performer_checkpoint",
|
| 1179 |
type=str,
|
| 1180 |
default=str(WEIGHTS_DIR / "BioMod.pth"),
|
| 1181 |
+
dest='biomod_checkpoint',
|
| 1182 |
)
|
| 1183 |
parser.add_argument("--biomod_data_dir", "--module_data_dir", type=str, default=str(ASSET_DIR / "biomod_data_all"), dest='module_data_dir')
|
| 1184 |
parser.add_argument("--gene_list_path", type=str, default=str(ASSET_DIR / "selected_genes_27k.txt"))
|
BioMod/command.txt
CHANGED
|
@@ -28,4 +28,3 @@ torchrun --standalone --nproc_per_node="${NPROC_PER_NODE:-1}" \
|
|
| 28 |
--dec_embed_dim 1024
|
| 29 |
|
| 30 |
# Adjust cell count and optimization settings to your own dataset and experiment.
|
| 31 |
-
# Add -use_wandb --api_key "$WANDB_API_KEY" to enable optional experiment logging.
|
|
|
|
| 28 |
--dec_embed_dim 1024
|
| 29 |
|
| 30 |
# Adjust cell count and optimization settings to your own dataset and experiment.
|
|
|
BioMod/environment.yml
CHANGED
|
@@ -62,7 +62,6 @@ dependencies:
|
|
| 62 |
- sentence-transformers>=2.7.0
|
| 63 |
- transformers>=4.40.0
|
| 64 |
- openai>=1.40.0
|
| 65 |
-
- wandb==0.23.0
|
| 66 |
|
| 67 |
# Only needed if running legacy pickles on Python < 3.8.
|
| 68 |
- pickle5>=0.0.12; python_version < "3.8"
|
|
|
|
| 62 |
- sentence-transformers>=2.7.0
|
| 63 |
- transformers>=4.40.0
|
| 64 |
- openai>=1.40.0
|
|
|
|
| 65 |
|
| 66 |
# Only needed if running legacy pickles on Python < 3.8.
|
| 67 |
- pickle5>=0.0.12; python_version < "3.8"
|
BioMod/gene_perturbation/components/model.py
CHANGED
|
@@ -11,7 +11,7 @@
|
|
| 11 |
- Use attention pooling to aggregate genetic information within the module
|
| 12 |
|
| 13 |
2. BioMod: Encoder-only Transformer model
|
| 14 |
-
- Use
|
| 15 |
- Support Masked Autoencoder (MAE) pre-training paradigm
|
| 16 |
|
| 17 |
3. SCFM_BioMod: wrapper for pre-trained models
|
|
@@ -45,8 +45,8 @@
|
|
| 45 |
import scipy
|
| 46 |
|
| 47 |
# ============================================================
|
| 48 |
-
# Import
|
| 49 |
-
#
|
| 50 |
# ============================================================
|
| 51 |
try:
|
| 52 |
from BioMod.models_biomod import (
|
|
|
|
| 11 |
- Use attention pooling to aggregate genetic information within the module
|
| 12 |
|
| 13 |
2. BioMod: Encoder-only Transformer model
|
| 14 |
+
- Use the BioMod encoder to contextualize module tokens
|
| 15 |
- Support Masked Autoencoder (MAE) pre-training paradigm
|
| 16 |
|
| 17 |
3. SCFM_BioMod: wrapper for pre-trained models
|
|
|
|
| 45 |
import scipy
|
| 46 |
|
| 47 |
# ============================================================
|
| 48 |
+
# Import BioMod model components (does not depend on gears_reduce)
|
| 49 |
+
# Attention and encoder behavior are defined in BioMod.models_biomod.
|
| 50 |
# ============================================================
|
| 51 |
try:
|
| 52 |
from BioMod.models_biomod import (
|
BioMod/gene_perturbation/components/trainer.py
CHANGED
|
@@ -68,8 +68,7 @@ class GEARS:
|
|
| 68 |
[Core functions]
|
| 69 |
1. Manage the entire training process (data, model, optimization, evaluation)
|
| 70 |
2. Support distributed training (DDP)
|
| 71 |
-
3. Support
|
| 72 |
-
4. Support model saving/loading
|
| 73 |
|
| 74 |
[Usage process]
|
| 75 |
```python
|
|
@@ -86,7 +85,6 @@ def __init__(self, pert_data,
|
|
| 86 |
train_bs = 48,
|
| 87 |
test_bs = 48,
|
| 88 |
device = 'cuda',
|
| 89 |
-
weight_bias_track = True,
|
| 90 |
proj_name = 'GEARS',
|
| 91 |
exp_name = 'GEARS',):
|
| 92 |
"""
|
|
@@ -97,20 +95,12 @@ def __init__(self, pert_data,
|
|
| 97 |
2. Control group expression value (used for direction loss calculation)
|
| 98 |
3. dict_filter: Non-zero gene index corresponding to each perturbation (affects the training loss calculation range)
|
| 99 |
"""
|
| 100 |
-
self.weight_bias_track = weight_bias_track
|
| 101 |
self.is_master = is_master
|
| 102 |
self.local_rank = local_rank
|
| 103 |
self.world_size = world_size
|
| 104 |
self.proj_name = proj_name
|
| 105 |
self.exp_name = exp_name
|
| 106 |
|
| 107 |
-
# Weights & Biases Experiment Tracking
|
| 108 |
-
if self.weight_bias_track:
|
| 109 |
-
import wandb
|
| 110 |
-
wandb.init(project=proj_name, name=exp_name, entity = "deanghoss")
|
| 111 |
-
self.wandb = wandb
|
| 112 |
-
else:
|
| 113 |
-
self.wandb = None
|
| 114 |
|
| 115 |
self.device = f'cuda:{local_rank}'
|
| 116 |
torch.cuda.set_device(device)
|
|
@@ -309,8 +299,6 @@ def model_initialize(self, hidden_size = 64,
|
|
| 309 |
print('Use mode:',mode)
|
| 310 |
self.config.update(kwargs)
|
| 311 |
|
| 312 |
-
if self.wandb:
|
| 313 |
-
self.wandb.config.update(self.config)
|
| 314 |
|
| 315 |
# ============================================================
|
| 316 |
# [Key] Constructing a co-expression graph
|
|
@@ -634,8 +622,6 @@ def train(self,
|
|
| 634 |
|
| 635 |
if (step + 1) % 50 == 0:
|
| 636 |
if self.is_master:
|
| 637 |
-
if self.wandb:
|
| 638 |
-
self.wandb.log({'training_loss': loss.item()})
|
| 639 |
log = "Epoch {} Step {} Train Loss: {:.4f}"
|
| 640 |
print_sys(log.format(epoch + 1, step + 1, loss.item()))
|
| 641 |
dist.barrier()
|
|
@@ -655,10 +641,6 @@ def train(self,
|
|
| 655 |
|
| 656 |
if self.is_master:
|
| 657 |
print_sys(log.format(epoch + 1, val_metrics['mse'], val_metrics['mse_de']))
|
| 658 |
-
if self.wandb:
|
| 659 |
-
metrics = ['mse', 'pearson']
|
| 660 |
-
for m in metrics:
|
| 661 |
-
self.wandb.log({'val_'+m: val_metrics[m],'val_de_'+m: val_metrics[m + '_de']})
|
| 662 |
# Use `mse_de` as early stop/select optimal criterion: pay more attention to DE gene prediction quality.
|
| 663 |
if val_metrics['mse_de'] < min_val:
|
| 664 |
min_val = val_metrics['mse_de']
|
|
@@ -702,12 +684,6 @@ def train(self,
|
|
| 702 |
log = "Best performing model: Test Top 20 DE MSE: {:.4f}"
|
| 703 |
if self.is_master:
|
| 704 |
print_sys(log.format(test_metrics['mse_de']))
|
| 705 |
-
if self.wandb:
|
| 706 |
-
metrics = ['mse', 'pearson']
|
| 707 |
-
for m in metrics:
|
| 708 |
-
self.wandb.log({'test_' + m: test_metrics[m],
|
| 709 |
-
'test_de_'+m: test_metrics[m + '_de']
|
| 710 |
-
})
|
| 711 |
|
| 712 |
out = deeper_analysis(self.adata, test_res)
|
| 713 |
out_non_dropout = non_dropout_analysis(self.adata, test_res)
|
|
@@ -729,8 +705,6 @@ def train(self,
|
|
| 729 |
for name, result in subgroup_analysis.items():
|
| 730 |
for m in result.keys():
|
| 731 |
subgroup_analysis[name][m] = np.mean(subgroup_analysis[name][m])
|
| 732 |
-
if self.wandb:
|
| 733 |
-
self.wandb.log({'test_' + name + '_' + m: subgroup_analysis[name][m]})
|
| 734 |
|
| 735 |
print_sys('test_' + name + '_' + m + ': ' + str(subgroup_analysis[name][m]))
|
| 736 |
|
|
@@ -756,8 +730,6 @@ def train(self,
|
|
| 756 |
for name, result in subgroup_analysis.items():
|
| 757 |
for m in result.keys():
|
| 758 |
subgroup_analysis[name][m] = np.mean(subgroup_analysis[name][m])
|
| 759 |
-
if self.wandb:
|
| 760 |
-
self.wandb.log({'test_' + name + '_' + m: subgroup_analysis[name][m]})
|
| 761 |
|
| 762 |
print_sys('test_' + name + '_' + m + ': ' + str(subgroup_analysis[name][m]))
|
| 763 |
print_sys('Done!')
|
|
|
|
| 68 |
[Core functions]
|
| 69 |
1. Manage the entire training process (data, model, optimization, evaluation)
|
| 70 |
2. Support distributed training (DDP)
|
| 71 |
+
3. Support model saving/loading
|
|
|
|
| 72 |
|
| 73 |
[Usage process]
|
| 74 |
```python
|
|
|
|
| 85 |
train_bs = 48,
|
| 86 |
test_bs = 48,
|
| 87 |
device = 'cuda',
|
|
|
|
| 88 |
proj_name = 'GEARS',
|
| 89 |
exp_name = 'GEARS',):
|
| 90 |
"""
|
|
|
|
| 95 |
2. Control group expression value (used for direction loss calculation)
|
| 96 |
3. dict_filter: Non-zero gene index corresponding to each perturbation (affects the training loss calculation range)
|
| 97 |
"""
|
|
|
|
| 98 |
self.is_master = is_master
|
| 99 |
self.local_rank = local_rank
|
| 100 |
self.world_size = world_size
|
| 101 |
self.proj_name = proj_name
|
| 102 |
self.exp_name = exp_name
|
| 103 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
|
| 105 |
self.device = f'cuda:{local_rank}'
|
| 106 |
torch.cuda.set_device(device)
|
|
|
|
| 299 |
print('Use mode:',mode)
|
| 300 |
self.config.update(kwargs)
|
| 301 |
|
|
|
|
|
|
|
| 302 |
|
| 303 |
# ============================================================
|
| 304 |
# [Key] Constructing a co-expression graph
|
|
|
|
| 622 |
|
| 623 |
if (step + 1) % 50 == 0:
|
| 624 |
if self.is_master:
|
|
|
|
|
|
|
| 625 |
log = "Epoch {} Step {} Train Loss: {:.4f}"
|
| 626 |
print_sys(log.format(epoch + 1, step + 1, loss.item()))
|
| 627 |
dist.barrier()
|
|
|
|
| 641 |
|
| 642 |
if self.is_master:
|
| 643 |
print_sys(log.format(epoch + 1, val_metrics['mse'], val_metrics['mse_de']))
|
|
|
|
|
|
|
|
|
|
|
|
|
| 644 |
# Use `mse_de` as early stop/select optimal criterion: pay more attention to DE gene prediction quality.
|
| 645 |
if val_metrics['mse_de'] < min_val:
|
| 646 |
min_val = val_metrics['mse_de']
|
|
|
|
| 684 |
log = "Best performing model: Test Top 20 DE MSE: {:.4f}"
|
| 685 |
if self.is_master:
|
| 686 |
print_sys(log.format(test_metrics['mse_de']))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 687 |
|
| 688 |
out = deeper_analysis(self.adata, test_res)
|
| 689 |
out_non_dropout = non_dropout_analysis(self.adata, test_res)
|
|
|
|
| 705 |
for name, result in subgroup_analysis.items():
|
| 706 |
for m in result.keys():
|
| 707 |
subgroup_analysis[name][m] = np.mean(subgroup_analysis[name][m])
|
|
|
|
|
|
|
| 708 |
|
| 709 |
print_sys('test_' + name + '_' + m + ': ' + str(subgroup_analysis[name][m]))
|
| 710 |
|
|
|
|
| 730 |
for name, result in subgroup_analysis.items():
|
| 731 |
for m in result.keys():
|
| 732 |
subgroup_analysis[name][m] = np.mean(subgroup_analysis[name][m])
|
|
|
|
|
|
|
| 733 |
|
| 734 |
print_sys('test_' + name + '_' + m + ': ' + str(subgroup_analysis[name][m]))
|
| 735 |
print_sys('Done!')
|
BioMod/gene_perturbation/train.py
CHANGED
|
@@ -169,7 +169,6 @@ def main(parser):
|
|
| 169 |
local_rank=local_rank,
|
| 170 |
is_master=is_master,
|
| 171 |
world_size=world_size,
|
| 172 |
-
weight_bias_track=args.wandb,
|
| 173 |
train_bs=args.batch_size,
|
| 174 |
test_bs=args.test_batch_size,
|
| 175 |
device=device,
|
|
@@ -271,7 +270,6 @@ def main(parser):
|
|
| 271 |
parser.add_argument('--finetune_method', type=str, default=None)
|
| 272 |
parser.add_argument('--mode', type=str, default='v1')
|
| 273 |
parser.add_argument('--accumulation_steps', type=int, default=1)
|
| 274 |
-
parser.add_argument('-wandb', action='store_true', help='Enable Weights & Biases logging')
|
| 275 |
parser.add_argument('-record_pred', action='store_true', help='Record predictions for analysis')
|
| 276 |
|
| 277 |
# ============================================================
|
|
|
|
| 169 |
local_rank=local_rank,
|
| 170 |
is_master=is_master,
|
| 171 |
world_size=world_size,
|
|
|
|
| 172 |
train_bs=args.batch_size,
|
| 173 |
test_bs=args.test_batch_size,
|
| 174 |
device=device,
|
|
|
|
| 270 |
parser.add_argument('--finetune_method', type=str, default=None)
|
| 271 |
parser.add_argument('--mode', type=str, default='v1')
|
| 272 |
parser.add_argument('--accumulation_steps', type=int, default=1)
|
|
|
|
| 273 |
parser.add_argument('-record_pred', action='store_true', help='Record predictions for analysis')
|
| 274 |
|
| 275 |
# ============================================================
|
BioMod/gene_perturbation/tutorial.ipynb
CHANGED
|
@@ -139,7 +139,6 @@
|
|
| 139 |
"run_env.setdefault('NUMBA_NUM_THREADS', '1')\n",
|
| 140 |
"run_env['NUMBA_DISABLE_JIT'] = '1'\n",
|
| 141 |
"run_env.setdefault('NUMBA_CACHE_DIR', str(Path(tempfile.gettempdir()) / f'biomod_perturbation_{os.getpid()}'))\n",
|
| 142 |
-
"run_env['WANDB_MODE'] = 'offline'\n",
|
| 143 |
"run_env.setdefault('TORCH_NCCL_ASYNC_ERROR_HANDLING', '1')\n",
|
| 144 |
"run_env.pop('NCCL_ASYNC_ERROR_HANDLING', None)\n",
|
| 145 |
"run_env.setdefault('NCCL_DEBUG', 'WARN')\n",
|
|
|
|
| 139 |
"run_env.setdefault('NUMBA_NUM_THREADS', '1')\n",
|
| 140 |
"run_env['NUMBA_DISABLE_JIT'] = '1'\n",
|
| 141 |
"run_env.setdefault('NUMBA_CACHE_DIR', str(Path(tempfile.gettempdir()) / f'biomod_perturbation_{os.getpid()}'))\n",
|
|
|
|
| 142 |
"run_env.setdefault('TORCH_NCCL_ASYNC_ERROR_HANDLING', '1')\n",
|
| 143 |
"run_env.pop('NCCL_ASYNC_ERROR_HANDLING', None)\n",
|
| 144 |
"run_env.setdefault('NCCL_DEBUG', 'WARN')\n",
|
BioMod/models_biomod.py
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
BioMod/requirements.txt
CHANGED
|
@@ -58,7 +58,6 @@ lifelines>=0.27.0
|
|
| 58 |
sentence-transformers>=2.7.0
|
| 59 |
transformers>=4.40.0
|
| 60 |
openai>=1.40.0
|
| 61 |
-
wandb==0.23.0
|
| 62 |
|
| 63 |
# Only needed if running legacy pickles on Python < 3.8.
|
| 64 |
pickle5>=0.0.12; python_version < "3.8"
|
|
|
|
| 58 |
sentence-transformers>=2.7.0
|
| 59 |
transformers>=4.40.0
|
| 60 |
openai>=1.40.0
|
|
|
|
| 61 |
|
| 62 |
# Only needed if running legacy pickles on Python < 3.8.
|
| 63 |
pickle5>=0.0.12; python_version < "3.8"
|
data/downstream/final_data/drug_combination_synergy/MANIFEST.csv
CHANGED
|
@@ -27,4 +27,4 @@ data_summary/test_data/independent_set_newcombinatioin.csv,249186,0.2376,7b7aadb
|
|
| 27 |
data_summary/test_data/README.md,2680,0.0026,ae5fa6fe381c77e3e75411a556c43a8c2adbbff49354793f08baa9b17ec4d751
|
| 28 |
data_summary/test_data/readme.txt,210,0.0002,eb984e1c5f106090fbc532371bf950dfe7f54bd7045ea344209423c544128481
|
| 29 |
README.md,1469,0.0014,0f394dd952fcce76519f20f20ccb8ffbe37235a105ca5cf6becfe838dab4392b
|
| 30 |
-
SUMMARY.json,
|
|
|
|
| 27 |
data_summary/test_data/README.md,2680,0.0026,ae5fa6fe381c77e3e75411a556c43a8c2adbbff49354793f08baa9b17ec4d751
|
| 28 |
data_summary/test_data/readme.txt,210,0.0002,eb984e1c5f106090fbc532371bf950dfe7f54bd7045ea344209423c544128481
|
| 29 |
README.md,1469,0.0014,0f394dd952fcce76519f20f20ccb8ffbe37235a105ca5cf6becfe838dab4392b
|
| 30 |
+
SUMMARY.json,653,0.0006,c28072fba34d958ed863ab81aeb4b94a67f69c8c4d54b0f784f461986aed74a5
|
data/downstream/final_data/drug_combination_synergy/SUMMARY.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"runnable_data_path": "
|
| 3 |
"train_labels_shape": [
|
| 4 |
12415,
|
| 5 |
4
|
|
|
|
| 1 |
{
|
| 2 |
+
"runnable_data_path": "data/downstream/final_data/drug_combination_synergy/data_summary",
|
| 3 |
"train_labels_shape": [
|
| 4 |
12415,
|
| 5 |
4
|
data/downstream/final_data/single_drug_response/MANIFEST.csv
CHANGED
|
@@ -238,4 +238,4 @@ optional_reference/DeepCDR_with_mut_with_gexp_with_methy_256_256_256_bn_relu_GAP
|
|
| 238 |
optional_reference/GDSC/drug_graph_feat.zip,576374,0.5497,a059696577dcf04c37dbcd41ba67018fbe3ac800e450a275f0ed8201a012e4c2
|
| 239 |
optional_reference/Missing_data_pre/records_pre_all.txt,1257750,1.1995,9298474435977de7934303f73627078954e99e453ba0588029a0315da04c9c60
|
| 240 |
README.md,1443,0.0014,389085fc257709861b0cb3e48043058b8be8632b006ec69290d055af2fdfec99
|
| 241 |
-
SUMMARY.json,
|
|
|
|
| 238 |
optional_reference/GDSC/drug_graph_feat.zip,576374,0.5497,a059696577dcf04c37dbcd41ba67018fbe3ac800e450a275f0ed8201a012e4c2
|
| 239 |
optional_reference/Missing_data_pre/records_pre_all.txt,1257750,1.1995,9298474435977de7934303f73627078954e99e453ba0588029a0315da04c9c60
|
| 240 |
README.md,1443,0.0014,389085fc257709861b0cb3e48043058b8be8632b006ec69290d055af2fdfec99
|
| 241 |
+
SUMMARY.json,312,0.0003,3fc48a122addd1c117877d8bd3323c7ea8fe5df0fd13f7f732718c2f15e20865
|
data/downstream/final_data/single_drug_response/SUMMARY.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"runnable_data_path": "
|
| 3 |
"drug_info_shape": [
|
| 4 |
266,
|
| 5 |
8
|
|
|
|
| 1 |
{
|
| 2 |
+
"runnable_data_path": "data/downstream/final_data/single_drug_response/data",
|
| 3 |
"drug_info_shape": [
|
| 4 |
266,
|
| 5 |
8
|