1- import argparse
2- import asyncio
1+ from __future__ import annotations
2+
3+ from pathlib import Path
34
45import numpy as np
56import pandas as pd
1213
1314
1415# Load Proteomics
15- def load_proteomics_data ( datafilename , context_name ) :
16+ def process_proteomics_data ( path : Path ) -> pd . DataFrame :
1617 """Load proteomics data from a given context and filename."""
1718 config = Config ()
1819 data_path = config .data_dir / "data_matrices" / context_name / datafilename
@@ -45,7 +46,7 @@ def load_proteomics_data(datafilename, context_name):
4546
4647
4748# read map to convert to entrez
48- async def load_gene_symbol_map (gene_symbols : list [str ]):
49+ async def load_gene_symbol_map (gene_symbols : list [str ], entrez_map : Path | None = None ):
4950 """Add descirption...."""
5051 config = Config ()
5152 filepath = config .data_dir / "proteomics_entrez_map.csv"
@@ -62,7 +63,15 @@ async def load_gene_symbol_map(gene_symbols: list[str]):
6263 return df [~ df .index .duplicated ()]
6364
6465
65- def abundance_to_bool_group (context_name , group_name , abundance_matrix , rep_ratio , hi_rep_ratio , quantile ):
66+ def abundance_to_bool_group (
67+ context_name ,
68+ group_name ,
69+ abundance_matrix ,
70+ replicate_ratio ,
71+ high_confidence_replicate_ratio ,
72+ quantile ,
73+ output_boolean_filepath : Path ,
74+ ):
6675 """Descrioption...."""
6776 config = Config ()
6877 output_dir = config .result_dir / context_name / "proteomics"
@@ -156,11 +165,15 @@ def load_empty_dict():
156165
157166
158167async def proteomics_gen (
159- config_file : str ,
160- rep_ratio : float = 0.5 ,
161- group_ratio : float = 0.5 ,
162- hi_rep_ratio : float = 0.5 ,
163- hi_group_ratio : float = 0.5 ,
168+ context_name : str ,
169+ config_filepath : Path ,
170+ matrix_filepath : Path ,
171+ output_boolean_filepath : Path ,
172+ input_entrez_map : Path | None = None ,
173+ replicate_ratio : float = 0.5 ,
174+ batch_ratio : float = 0.5 ,
175+ high_confidence_replicate_ratio : float = 0.7 ,
176+ high_confience_batch_ratio : float = 0.7 ,
164177 quantile : int = 25 ,
165178):
166179 """Generate proteomics data."""
0 commit comments