comparison decoupler_aucell_score.py @ 3:4fa5f370599f draft

planemo upload for repository https://github.com/ebi-gene-expression-group/container-galaxy-sc-tertiary/ commit c8c39f14eeee6e7a6d097fd0cb9430b12793eb8b
author ebi-gxa
date Thu, 09 Nov 2023 11:35:57 +0000
parents
children f321c60167d4
comparison
equal deleted inserted replaced
2:130e25d3ce92 3:4fa5f370599f
1 import argparse
2 import os
3 import tempfile
4
5 import anndata
6 import decoupler as dc
7 import pandas as pd
8
9
10 def read_gmt(gmt_file):
11 """
12 Reads a GMT file into a Pandas DataFrame.
13
14 Parameters
15 ----------
16 gmt_file : str
17 Path to the GMT file.
18
19 Returns
20 -------
21 pd.DataFrame
22 A DataFrame with the gene sets. Each row represents a gene set, and the columns are "gene_set_name", "gene_set_url", and "genes".
23 >>> line = "HALLMARK_NOTCH_SIGNALING\\thttp://www.gsea-msigdb.org/gsea/msigdb/human/geneset/HALLMARK_NOTCH_SIGNALING\\tJAG1\\tNOTCH3\\tNOTCH2\\tAPH1A\\tHES1\\tCCND1\\tFZD1\\tPSEN2\\tFZD7\\tDTX1\\tDLL1\\tFZD5\\tMAML2\\tNOTCH1\\tPSENEN\\tWNT5A\\tCUL1\\tWNT2\\tDTX4\\tSAP30\\tPPARD\\tKAT2A\\tHEYL\\tSKP1\\tRBX1\\tTCF7L2\\tARRB1\\tLFNG\\tPRKCA\\tDTX2\\tST3GAL6\\tFBXW11\\n"
24 >>> line2 = "HALLMARK_APICAL_SURFACE\\thttp://www.gsea-msigdb.org/gsea/msigdb/human/geneset/HALLMARK_APICAL_SURFACE\\tB4GALT1\\tRHCG\\tMAL\\tLYPD3\\tPKHD1\\tATP6V0A4\\tCRYBG1\\tSHROOM2\\tSRPX\\tMDGA1\\tTMEM8B\\tTHY1\\tPCSK9\\tEPHB4\\tDCBLD2\\tGHRL\\tLYN\\tGAS1\\tFLOT2\\tPLAUR\\tAKAP7\\tATP8B1\\tEFNA5\\tSLC34A3\\tAPP\\tGSTM3\\tHSPB1\\tSLC2A4\\tIL2RB\\tRTN4RL1\\tNCOA6\\tSULF2\\tADAM10\\tBRCA1\\tGATA3\\tAFAP1L2\\tIL2RG\\tCD160\\tADIPOR2\\tSLC22A12\\tNTNG1\\tSCUBE1\\tCX3CL1\\tCROCC\\n"
25 >>> temp_dir = tempfile.gettempdir()
26 >>> temp_gmt = os.path.join(temp_dir, "temp_file.gmt")
27 >>> with open(temp_gmt, "w") as f:
28 ... f.write(line)
29 ... f.write(line2)
30 288
31 380
32 >>> df = read_gmt(temp_gmt)
33 >>> df.shape[0]
34 2
35 >>> df.columns == ["gene_set_name", "genes"]
36 array([ True, True])
37 >>> df.loc[df["gene_set_name"] == "HALLMARK_APICAL_SURFACE"].genes.tolist()[0].startswith("B4GALT1")
38 True
39 """
40 # Read the GMT file into a list of lines
41 with open(gmt_file, "r") as f:
42 lines = f.readlines()
43
44 # Create a list of dictionaries, where each dictionary represents a gene set
45 gene_sets = []
46 for line in lines:
47 fields = line.strip().split("\t")
48 gene_set = {"gene_set_name": fields[0], "genes": ",".join(fields[2:])}
49 gene_sets.append(gene_set)
50
51 # Convert the list of dictionaries to a DataFrame
52 return pd.DataFrame(gene_sets)
53
54
55 def score_genes_aucell(
56 adata: anndata.AnnData, gene_list: list, score_name: str, use_raw=False
57 ):
58 """Score genes using Aucell.
59
60 Parameters
61 ----------
62 adata : anndata.AnnData
63 gene_list : list
64 score_names : str
65 use_raw : bool, optional
66
67 >>> import scanpy as sc
68 >>> import decoupler as dc
69 >>> adata = sc.datasets.pbmc68k_reduced()
70 >>> gene_list = adata.var[adata.var.index.str.startswith("RP")].index.tolist()
71 >>> score_genes_aucell(adata, gene_list, "ribosomal_aucell", use_raw=False)
72 >>> "ribosomal_aucell" in adata.obs.columns
73 True
74 """
75 # make a data.frame with two columns, geneset and gene_id, geneset filled with score_names and gene_id with gene_list, one row per element
76 geneset_df = pd.DataFrame(
77 {
78 "gene_id": gene_list,
79 "geneset": score_name,
80 }
81 )
82 # run decoupler's run_aucell
83 dc.run_aucell(
84 adata, net=geneset_df, source="geneset", target="gene_id", use_raw=use_raw
85 )
86 # copy .obsm['aucell_estimate'] matrix columns to adata.obs using the column names
87 adata.obs[score_name] = adata.obsm["aucell_estimate"][score_name]
88
89
90 def run_for_genelists(
91 adata, gene_lists, score_names, use_raw=False, gene_symbols_field="gene_symbols"
92 ):
93 if len(gene_lists) == len(score_names):
94 for gene_list, score_names in zip(gene_lists, score_names):
95 genes = gene_list.split(",")
96 ens_gene_ids = adata.var[adata.var[gene_symbols_field].isin(genes)].index
97 score_genes_aucell(
98 adata,
99 ens_gene_ids,
100 f"AUCell_{score_names}",
101 use_raw,
102 )
103 else:
104 raise ValueError(
105 "The number of gene lists (separated by :) and score names (separated by :) must be the same"
106 )
107
108
109 if __name__ == "__main__":
110 # Create command-line arguments parser
111 parser = argparse.ArgumentParser(description="Score genes using Aucell")
112 parser.add_argument("--input_file", type=str, help="Path to input AnnData file")
113 parser.add_argument("--output_file", type=str, help="Path to output file")
114 parser.add_argument("--gmt_file", type=str, help="Path to GMT file", required=False)
115 # add argument for gene sets to score
116 parser.add_argument(
117 "--gene_sets_to_score",
118 type=str,
119 required=False,
120 help="Comma separated list of gene sets to score (the need to be in the gmt file)",
121 )
122 # add argument for gene list (comma separated) to score
123 parser.add_argument(
124 "--gene_lists_to_score",
125 type=str,
126 required=False,
127 help="Comma separated list of genes to score. You can have more than one set of genes, separated by colon :",
128 )
129 # argument for the score name when using the gene list
130 parser.add_argument(
131 "--score_names",
132 type=str,
133 required=False,
134 help="Name of the score column when using the gene list. You can have more than one set of score names, separated by colon :. It should be the same length as the number of gene lists.",
135 )
136 parser.add_argument(
137 "--gene_symbols_field",
138 type=str,
139 help="Name of the gene symbols field in the AnnData object",
140 )
141 parser.add_argument("--use_raw", action="store_true", help="Use raw data")
142 parser.add_argument(
143 "--write_anndata", action="store_true", help="Write the modified AnnData object"
144 )
145
146 # Parse command-line arguments
147 args = parser.parse_args()
148
149 # Load input AnnData object
150 adata = anndata.read_h5ad(args.input_file)
151
152 if args.gene_sets_to_score is not None and args.gmt_file is not None:
153 # Load MSigDB file in GMT format
154 msigdb = read_gmt(args.gmt_file)
155
156 gene_sets_to_score = args.gene_sets_to_score.split(",")
157 # Score genes by their ensembl ids using the score_genes_aucell function
158 for _, row in msigdb.iterrows():
159 gene_set_name = row["gene_set_name"]
160 if gene_set_name in gene_sets_to_score:
161 genes = row["genes"].split(",")
162 # Convert gene symbols to ensembl ids by using the columns gene_symbols and index in adata.var specific to the gene set
163 ens_gene_ids = adata.var[
164 adata.var[args.gene_symbols_field].isin(genes)
165 ].index
166 score_genes_aucell(
167 adata, ens_gene_ids, f"AUCell_{gene_set_name}", args.use_raw
168 )
169 elif args.gene_lists_to_score is not None and args.score_names is not None:
170 gene_lists = args.gene_lists_to_score.split(":")
171 score_names = args.score_names.split(",")
172 run_for_genelists(
173 adata, gene_lists, score_names, args.use_raw, args.gene_symbols_field
174 )
175
176 # Save the modified AnnData object or generate a file with cells as rows and the new score_names columns
177 if args.write_anndata:
178 adata.write_h5ad(args.output_file)
179 else:
180 new_columns = [col for col in adata.obs.columns if col.startswith("AUCell_")]
181 adata.obs[new_columns].to_csv(args.output_file, sep="\t", index=True)