File size: 10,468 Bytes
b61fce2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 | #!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""CLI entry for DE/DIR pipeline stages."""
import argparse
import sys
from stages.prepare import process_cell_line
from stages.retrieve import build_retrieval_results
from stages.prompt import generate_prompts
from stages.infer import run_inference
from stages.infer_api import run_inference_api
from stages.single_case.prompt import generate_single_case_prompt
def _add_prepare_args(p: argparse.ArgumentParser) -> None:
p.add_argument("--h5ad", required=True, help="Input .h5ad file")
p.add_argument("--out-dir", required=True, help="Output directory for CSVs")
p.add_argument("--cell-line", required=True, help="Cell line name for output file prefix")
p.add_argument("--perturbation-col", default="drug", help="AnnData obs column for perturbation")
p.add_argument("--control-value", default="DMSO_TF", help="Control group name")
p.add_argument("--train-fraction", type=float, default=0.3, help="Train split fraction")
p.add_argument("--seed", type=int, default=42)
p.add_argument("--fdr", type=float, default=0.05)
p.add_argument("--lfc", type=float, default=0.25)
p.add_argument("--pval-neg", type=float, default=0.1)
p.add_argument("--n-neg", type=int, default=200)
def _add_retrieve_args(p: argparse.ArgumentParser) -> None:
p.add_argument("--data-csv", required=True, help="Input CSV with pert/gene/label/split")
p.add_argument("--drug-sim", required=True, help="Drug similarity JSON")
p.add_argument("--gene-sim", required=True, help="Gene similarity JSON (required)")
p.add_argument("--out", required=True, help="Output retrieval JSON")
p.add_argument("--budget", type=int, default=10)
p.add_argument("--seed", type=int, default=0)
p.add_argument("--max-cases", type=int, default=None)
p.add_argument("--case-split", default="test", choices=["train", "test"], help="Which split to generate retrieval for")
def _add_prompt_args(p: argparse.ArgumentParser) -> None:
p.add_argument("--retrieval", required=True, help="Retrieval JSON from previous stage")
p.add_argument("--drug-desc", required=True, help="Drug description JSON")
p.add_argument("--gene-desc", required=True, help="Gene description JSON")
p.add_argument("--template", default=None, help="Prompt template file (optional)")
p.add_argument("--out", required=True, help="Output prompts text file")
p.add_argument("--cell-line-idx", type=int, default=None)
p.add_argument("--max-cases", type=int, default=None)
p.add_argument("--seed", type=int, default=42)
def _add_infer_args(p: argparse.ArgumentParser) -> None:
p.add_argument("--model", required=True, help="HF model path or name")
p.add_argument("--prompts", required=True, help="Prompt text file")
p.add_argument("--out", required=True, help="Output text file")
p.add_argument("--batch-size", type=int, default=4)
p.add_argument("--max-new-tokens", type=int, default=512)
p.add_argument("--temperature", type=float, default=0.6)
p.add_argument("--top-p", type=float, default=0.9)
p.add_argument("--dtype", default="bfloat16", choices=["bfloat16", "float16", "float32"])
p.add_argument("--device-map", default="auto")
p.add_argument("--chat-template", default=None, help="Optional chat template file to override tokenizer.chat_template")
def _add_infer_api_args(p: argparse.ArgumentParser) -> None:
p.add_argument("--api-url", required=True, help="API endpoint URL")
p.add_argument("--api-model", required=True, help="API model name")
p.add_argument("--api-key", default=None, help="API key (or set LLM_DRUG_API_KEY)")
p.add_argument("--prompts", required=True, help="Prompt text file")
p.add_argument("--out", required=True, help="Output text file")
p.add_argument("--max-new-tokens", type=int, default=512)
p.add_argument("--temperature", type=float, default=0.6)
p.add_argument("--top-p", type=float, default=0.9)
p.add_argument("--timeout", type=int, default=60)
p.add_argument("--sleep-secs", type=float, default=0.0)
def build_parser() -> argparse.ArgumentParser:
parser = argparse.ArgumentParser(description="LLM Drug pipeline (DE/DIR) CLI")
task_sub = parser.add_subparsers(dest="task", required=True)
for task_name in ("de", "dir"):
task_parser = task_sub.add_parser(task_name)
stage_sub = task_parser.add_subparsers(dest="stage", required=True)
prepare_p = stage_sub.add_parser("prepare")
_add_prepare_args(prepare_p)
retrieve_p = stage_sub.add_parser("retrieve")
_add_retrieve_args(retrieve_p)
prompt_p = stage_sub.add_parser("prompt")
_add_prompt_args(prompt_p)
infer_p = stage_sub.add_parser("infer")
_add_infer_args(infer_p)
infer_api_p = stage_sub.add_parser("infer-api")
_add_infer_api_args(infer_api_p)
single_p = task_sub.add_parser("single")
single_stage = single_p.add_subparsers(dest="stage", required=True)
single_prompt = single_stage.add_parser("prompt")
single_prompt.add_argument("--pert", required=True, help="Drug/perturbation name")
single_prompt.add_argument("--gene", required=True, help="Gene name")
single_prompt.add_argument("--cell-line", required=True, help="Cell line name (must exist in template)")
single_prompt.add_argument("--mode", default="de", choices=["de", "dir"], help="Prompt type")
single_prompt.add_argument("--data-csv", required=True, help="CSV with pert/gene/label/split for retrieval")
single_prompt.add_argument("--drug-desc", required=True, help="Drug description JSON")
single_prompt.add_argument("--gene-desc", required=True, help="Gene description JSON")
single_prompt.add_argument("--drug-sim", required=True, help="Drug similarity JSON")
single_prompt.add_argument("--gene-sim", required=True, help="Gene similarity JSON")
single_prompt.add_argument("--template", default=None, help="Prompt template file (optional)")
single_prompt.add_argument("--out", required=True, help="Output prompt text file")
single_prompt.add_argument("--max-candidates", type=int, default=10)
single_prompt.add_argument("--budget", type=int, default=10)
single_prompt.add_argument("--case-split", default="train", choices=["train", "test", "all"])
single_prompt.add_argument("--seed", type=int, default=42)
single_prompt.add_argument("--llm-api-url", default=None, help="LLM API endpoint for fallback similarity")
single_prompt.add_argument("--llm-api-model", default=None, help="LLM model name for fallback similarity")
single_prompt.add_argument("--llm-api-key", default=None, help="API key (or set LLM_DRUG_API_KEY)")
single_prompt.add_argument("--llm-candidate-pool", type=int, default=80)
single_prompt.add_argument("--llm-timeout", type=int, default=60)
return parser
def main(argv: list[str]) -> int:
parser = build_parser()
args = parser.parse_args(argv)
if args.task in ("de", "dir") and args.stage == "prepare":
process_cell_line(
adata_path=args.h5ad,
output_dir=args.out_dir,
cell_line_name=args.cell_line,
perturbation_col=args.perturbation_col,
control_value=args.control_value,
train_fraction=args.train_fraction,
seed=args.seed,
fdr=args.fdr,
lfc=args.lfc,
pval_neg=args.pval_neg,
n_neg=args.n_neg,
)
return 0
if args.task in ("de", "dir") and args.stage == "retrieve":
build_retrieval_results(
data_csv=args.data_csv,
drug_sim_json=args.drug_sim,
out_json=args.out,
gene_sim_json=args.gene_sim,
budget=args.budget,
seed=args.seed,
max_cases=args.max_cases,
case_split=args.case_split,
)
return 0
if args.task in ("de", "dir") and args.stage == "prompt":
generate_prompts(
task=args.task,
retrieval_json=args.retrieval,
drug_desc_json=args.drug_desc,
gene_desc_json=args.gene_desc,
template_file=args.template,
output_file=args.out,
cell_line_idx=args.cell_line_idx,
max_cases=args.max_cases,
seed=args.seed,
)
return 0
if args.task in ("de", "dir") and args.stage == "infer":
run_inference(
model_name=args.model,
prompts_file=args.prompts,
output_file=args.out,
batch_size=args.batch_size,
max_new_tokens=args.max_new_tokens,
temperature=args.temperature,
top_p=args.top_p,
dtype=args.dtype,
device_map=args.device_map,
chat_template_path=args.chat_template,
)
return 0
if args.task in ("de", "dir") and args.stage == "infer-api":
run_inference_api(
api_url=args.api_url,
api_model=args.api_model,
api_key=args.api_key,
prompts_file=args.prompts,
output_file=args.out,
max_new_tokens=args.max_new_tokens,
temperature=args.temperature,
top_p=args.top_p,
timeout=args.timeout,
sleep_secs=args.sleep_secs,
)
return 0
if args.task == "single" and args.stage == "prompt":
case_split = "" if args.case_split == "all" else args.case_split
generate_single_case_prompt(
task=args.mode,
pert=args.pert,
gene=args.gene,
cell_line=args.cell_line,
data_csv=args.data_csv,
drug_desc_json=args.drug_desc,
gene_desc_json=args.gene_desc,
drug_sim_json=args.drug_sim,
gene_sim_json=args.gene_sim,
template_file=args.template,
output_file=args.out,
max_candidates=args.max_candidates,
budget=args.budget,
case_split=case_split,
seed=args.seed,
llm_api_url=args.llm_api_url,
llm_api_model=args.llm_api_model,
llm_api_key=args.llm_api_key,
llm_candidate_pool=args.llm_candidate_pool,
llm_timeout=args.llm_timeout,
)
return 0
parser.error("Unknown stage")
return 2
if __name__ == "__main__":
raise SystemExit(main(sys.argv[1:]))
|