celescope 2.2.2__tar.gz → 2.3.0.dev0__tar.gz

{celescope-2.2.2/celescope.egg-info → celescope-2.3.0.dev0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: celescope
-Version: 2.2.2
+Version: 2.3.0.dev0
 Summary: Single Cell Analysis Pipelines
 Home-page: https://github.com/singleron-RD/CeleScope
 Author: zhouyiqi

{celescope-2.2.2 → celescope-2.3.0.dev0}/celescope/__init__.py

@@ -1,6 +1,6 @@
 import os
 
-__VERSION__ = "2.2.2"
+__VERSION__ = "2.3.0dev"
 __version__ = __VERSION__
 
 ASSAY_LIST = [
@@ -58,7 +58,7 @@ HELP_DICT = {
     "genomeSAindexNbases": """For small genomes, the parameter --genomeSAindexNbases must to be scaled down, with a typical 
 value of min(14, log2(GenomeLength)/2 - 1). For example, for 1 megaBase genome, this is equal 
 to 9, for 100 kiloBase genome, this is equal to 7.""",
-    "chemistry": "`--chemistry auto` can auto-detect scopeV2 mRNA, scopeV3 mRNA, full length VDJ mRNA(flv_rna) and full length VDJ(flv). You need to explicitly use `--chemistry scopeV1` for legacy chemistry scopeV1. `--chemistry customized` is used for user defined combinations that you need to provide `--pattern`, `--whitelist` and `--linker` at the same time.",
+    "chemistry": "`--chemistry auto` can auto-detect most chemistry. You need to explicitly use `--chemistry GEXSCOPE-MicroBead` for legacy chemistry GEXSCOPE-MicroBead as it has no barcode whitelists. `--chemistry customized` is used for user defined combinations that you need to provide `--pattern` and optional `--whitelist` at the same time.",
 }
 
 # report metrics help

celescope-2.3.0.dev0/celescope/bulk_rna/__init__.py

@@ -0,0 +1,5 @@
+STEPS = ["sample", "starsolo", "cells", "split_fastq"]
+__ASSAY__ = "bulk_rna"
+REMOVE_FROM_MULTI = {
+    "cells",
+}

celescope-2.3.0.dev0/celescope/bulk_rna/multi_bulk_rna.py

@@ -0,0 +1,33 @@
+from celescope.bulk_rna.__init__ import __ASSAY__
+from celescope.tools.multi import Multi
+from celescope.tools.__init__ import (
+    OUTS_DIR,
+)
+
+
+class Multi_bulk_rna(Multi):
+    def starsolo(self, sample):
+        step = "starsolo"
+        arr = self.fq_dict[sample]
+        cmd_line = self.get_cmd_line(step, sample)
+        cmd = f'{cmd_line} ' f'--fq1 {arr["fq1_str"]} --fq2 {arr["fq2_str"]} '
+        self.process_cmd(cmd, step, sample, m=self.args.starMem, x=self.args.thread)
+
+    def split_fastq(self, sample):
+        step = "split_fastq"
+        arr = self.fq_dict[sample]
+        cmd_line = self.get_cmd_line(step, sample)
+        cmd = (
+            f'{cmd_line} '
+            f'--fq1 {arr["fq1_str"]} --fq2 {arr["fq2_str"]} --barcodes {self.outdir_dic[sample][OUTS_DIR]}/filtered/barcodes.tsv.gz'
+        )
+        self.process_cmd(cmd, step, sample, m=5, x=1)
+
+
+def main():
+    multi = Multi_bulk_rna(__ASSAY__)
+    multi.run()
+
+
+if __name__ == "__main__":
+    main()

celescope-2.3.0.dev0/celescope/bulk_rna/starsolo.py

@@ -0,0 +1,135 @@
+import subprocess
+import sys
+from celescope.tools.starsolo import (
+    Starsolo as tools_Starsolo,
+    create_solo_args,
+    Mapping,
+    Cells,
+    Demultiplexing,
+)
+from celescope.tools.step import s_common
+from celescope.__init__ import HELP_DICT
+from celescope.chemistry_dict import chemistry_dict
+
+SAM_attributes = "NH HI nM AS CR UR CB UB GX GN "
+
+
+class Starsolo(tools_Starsolo):
+    def __init__(self, args, display_title=None):
+        super().__init__(args, display_title=display_title)
+
+    def run_starsolo(self):
+        cmd = create_solo_args(
+            pattern_args=self.pattern_args,
+            whitelist_args=self.whitelist_args,
+            outFileNamePrefix=self.out_prefix + "_",
+            fq1=self.args.fq1,
+            fq2=self.args.fq2,
+            genomeDir=self.args.genomeDir,
+            soloCellFilter=self.args.soloCellFilter,
+            runThreadN=self.args.thread,
+            clip3pAdapterSeq=self.args.adapter_3p,
+            outFilterMatchNmin=self.args.outFilterMatchNmin,
+            soloFeatures=self.args.soloFeatures,
+            outSAMattributes=self.outSAMattributes,
+            soloCBmatchWLtype=self.args.soloCBmatchWLtype,
+            extra_starsolo_args=self.extra_starsolo_args,
+        )
+        if self.chemistry == "bulk_rna-bulk_vdj_match":
+            cmd += "--soloStrand Reverse \\\n"
+        sys.stderr.write(cmd)
+        subprocess.check_call(cmd, shell=True)
+        cmd = f"chmod -R 755 {self.solo_out_dir}"
+        sys.stderr.write(cmd)
+        subprocess.check_call(cmd, shell=True)
+
+
+def starsolo(args):
+    with Starsolo(args) as runner:
+        q30_cb, q30_umi, chemistry = runner.run()
+
+    with Mapping(args) as runner:
+        valid_reads, corrected = runner.run()
+
+    with Cells(args) as runner:
+        n_reads, q30_RNA = runner.run(chemistry, valid_reads)
+
+    with Demultiplexing(args) as runner:
+        runner.run(valid_reads, n_reads, corrected, q30_cb, q30_umi, q30_RNA)
+
+
+def get_opts_starsolo(parser, sub_program=True):
+    parser.add_argument(
+        "--chemistry",
+        help=HELP_DICT["chemistry"],
+        choices=list(chemistry_dict.keys()),
+        default="auto",
+    )
+    parser.add_argument(
+        "--pattern",
+        help="""The pattern of R1 reads, e.g. `C8L16C8L16C8L1U12T18`. The number after the letter represents the number 
+        of bases.  
+        - `C`: cell barcode  
+        - `L`: linker(common sequences)  
+        - `U`: UMI    
+        - `T`: poly T""",
+    )
+    parser.add_argument(
+        "--whitelist",
+        help="Cell barcode whitelist file path, one cell barcode per line. Multiple whitelist files are seperated by space.",
+    )
+    parser.add_argument(
+        "--adapter_3p",
+        help="Adapter sequence to clip from 3 prime. Multiple sequences are seperated by space",
+        default="AAAAAAAAAAAA",
+    )
+    parser.add_argument(
+        "--genomeDir",
+        help=HELP_DICT["genomeDir"],
+    )
+    parser.add_argument(
+        "--outFilterMatchNmin",
+        help="""Alignment will be output only if the number of matched bases 
+is higher than or equal to this value.""",
+        default=50,
+    )
+    # 96 wells
+    parser.add_argument(
+        "--soloCellFilter",
+        help="The same as the argument in STARsolo",
+        default="CellRanger2.2 384 0.99 10",
+    )
+    parser.add_argument(
+        "--starMem", help="Maximum memory that STAR can use.", default=32
+    )
+    parser.add_argument("--STAR_param", help=HELP_DICT["additional_param"], default="")
+    parser.add_argument(
+        "--SAM_attributes",
+        help=f"Additional attributes(other than {SAM_attributes}) to be added to SAM file",
+        default="",
+    )
+    parser.add_argument(
+        "--soloFeatures",
+        help="The same as the argument in STARsolo",
+        default="GeneFull_Ex50pAS Gene",
+    )
+    parser.add_argument(
+        "--soloCBmatchWLtype",
+        help="The same as the argument in STARsolo. Please note `EditDist 2` only works with `--soloType CB UMI Complex`. ",
+        default="1MM",
+    )
+
+    if sub_program:
+        parser.add_argument(
+            "--fq1",
+            help="R1 fastq file. Multiple files are separated by comma.",
+            required=True,
+        )
+        parser.add_argument(
+            "--fq2",
+            help="R2 fastq file. Multiple files are separated by comma.",
+            required=True,
+        )
+        parser = s_common(parser)
+
+    return parser

celescope-2.3.0.dev0/celescope/chemistry_dict.py

@@ -0,0 +1,81 @@
+from importlib import resources
+
+chemistry_dict = {
+    "auto": {
+        "pattern": ""  # auto detect
+    },
+    "customized": {
+        "pattern": ""  # user defined
+    },
+    "GEXSCOPE-MicroBead": {"pattern": "C12U8"},
+    "GEXSCOPE-V1": {
+        "pattern": "C8L16C8L16C8L1U12",
+        "bc": ["bc.txt", "bc.txt", "bc.txt"],
+    },
+    "GEXSCOPE-V2": {
+        "pattern": "C9L16C9L16C9L1U12",
+        "bc": ["bc1.txt", "bc2.txt", "bc3.txt"],
+    },
+    "GEXSCOPE-V3": {
+        # 0-3 bases offset at the begining
+        "pattern": "C9L6C9L6C9L1U12",
+        "bc": ["bc1.txt", "bc2.txt", "bc3.txt"],
+        "linker": ["linker1.txt", "linker2.txt"],
+    },
+    "flv_rna": {
+        "pattern": "C8L16C8L16C8U9L6",
+        "bc": ["bc.txt", "bc.txt", "bc.txt"],
+    },
+    "flv": {
+        "pattern": "U9C8L16C8L16C8",
+        "bc": ["bc.txt", "bc.txt", "bc.txt"],
+    },
+    "rna_5p-1": {
+        "pattern": "U8C9L4C9L4C9",
+    },
+    "rna_3p-1": {
+        "pattern": "C9L4C9L4C9U8",
+    },
+    "5p3p-1": {
+        "pattern": "C9C9C9U8",  # converted
+        "bc": ["bc1.txt", "bc2.txt", "bc3.txt"],
+    },
+    "rna_5p-2": {
+        "pattern": "C8L4U9C8L4C8",
+    },
+    "rna_3p-2": {
+        "pattern": "C8L4C8U9L4C8",
+    },
+    "5p3p-2": {
+        "pattern": "C8C8C8U9",  # converted
+        "bc": ["bc1.txt", "bc2.txt", "bc3.txt"],
+    },
+    "rna_5p-3": {
+        "pattern": "C9L6U9C9L4C9",
+    },
+    "rna_3p-3": {
+        "pattern": "C9L4C9U9L6C9",
+    },
+    "5p3p-3": {
+        "pattern": "C9C9C9U9",  # converted
+        "bc": ["bc1.txt", "bc2.txt", "bc3.txt"],
+    },
+    "bulk_vdj": {
+        "pattern": "L18C6U16",
+        "bc": ["bc.txt"],
+    },
+    "bulk_rna-V1": {
+        "pattern": "C9U12",
+        "bc": ["bc.txt"],
+    },
+    "bulk_rna-V2": {
+        "pattern": "L9C9U12",
+        "bc": ["bc.txt"],
+    },
+    "bulk_rna-bulk_vdj_match": {
+        "pattern": "L18C6U16",
+        "bc": ["bc.txt"],
+    },
+}
+
+chemistry_dir = str(resources.files("celescope.data.chemistry"))

celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V1/bc.txt

@@ -0,0 +1,96 @@
+AACGTGAT
+AAACATCG
+ATGCCTAA
+AGTGGTCA
+ACCACTGT
+ACATTGGC
+CAGATCTG
+CATCAAGT
+CGCTGATC
+ACAAGCTA
+CTGTAGCC
+AGTACAAG
+AACAACCA
+AACCGAGA
+AACGCTTA
+AAGACGGA
+AAGGTACA
+ACACAGAA
+ACAGCAGA
+ACCTCCAA
+ACGCTCGA
+ACGTATCA
+ACTATGCA
+AGAGTCAA
+AGATCGCA
+AGCAGGAA
+AGTCACTA
+ATCCTGTA
+ATTGAGGA
+CAACCACA
+GACTAGTA
+CAATGGAA
+CACTTCGA
+CAGCGTTA
+CATACCAA
+CCAGTTCA
+CCGAAGTA
+CCGTGAGA
+CCTCCTGA
+CGAACTTA
+CGACTGGA
+CGCATACA
+CTCAATGA
+CTGAGCCA
+CTGGCATA
+GAATCTGA
+CAAGACTA
+GAGCTGAA
+GATAGACA
+GCCACATA
+GCGAGTAA
+GCTAACGA
+GCTCGGTA
+GGAGAACA
+GGTGCGAA
+GTACGCAA
+GTCGTAGA
+GTCTGTCA
+GTGTTCTA
+TAGGATGA
+TATCAGCA
+TCCGTCTA
+TCTTCACA
+TGAAGAGA
+TGGAACAA
+TGGCTTCA
+TGGTGGTA
+TTCACGCA
+AACTCACC
+AAGAGATC
+AAGGACAC
+AATCCGTC
+AATGTTGC
+ACACGACC
+ACAGATTC
+AGATGTAC
+AGCACCTC
+AGCCATGC
+AGGCTAAC
+ATAGCGAC
+ATCATTCC
+ATTGGCTC
+CAAGGAGC
+CACCTTAC
+CCATCCTC
+CCGACAAC
+CCTAATCC
+CCTCTATC
+CGACACAC
+CGGATTGC
+CTAAGGTC
+GAACAGGC
+GACAGTGC
+GAGTTAGC
+GATGAATC
+GCCAAGAC

celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V1/linker1.txt

@@ -0,0 +1,4 @@
+ATCCACGTGCTTGAGA
+TCGGTGACAGCCATAT
+CGAACATGTAGGTCTC
+GATTGTCACTAACGCG

celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V1/linker2.txt

@@ -0,0 +1,4 @@
+TCAGCATGCGGCTACG
+CGTAGTCAGAAGCTGA
+GACTACGTATTAGCAT
+ATGCTGACTCCTAGTC

celescope-2.2.2/celescope/data/chemistry/scopeV3.0.1/bclist2 → celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V2/bc2.txt

@@ -93,4 +93,4 @@ GATAACCGC
 GCTGCGATA
 TACTCACCA
 ACCAGGTCA
-GCAACTTCA
+GCAACTTCA

celescope-2.2.2/celescope/data/chemistry/scopeV3.0.1/bclist3 → celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V2/bc3.txt

@@ -189,4 +189,4 @@ CTAGGTTGC
 AGGCAGAAC
 CTACTATGG
 CTCAATCGT
-GTAACGCTT
+GTAACGCTT

celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V2/linker1.txt

@@ -0,0 +1,4 @@
+ATCCACGTGCTTGAGA
+TCGGTGACAGCCATAT
+CGAACATGTAGGTCTC
+GATTGTCACTAACGCG

celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V2/linker2.txt

@@ -0,0 +1,4 @@
+TCAGCATGCGGCTACG
+CGTAGTCAGAAGCTGA
+GACTACGTATTAGCAT
+ATGCTGACTCCTAGTC

celescope-2.2.2/celescope/data/chemistry/scopeV3.0.1/bclist1 → celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V3/bc1.txt

@@ -93,4 +93,4 @@ TACCGTCTG
 TCTTGCTTG
 CCAGCTAAC
 CTTACGCAG
-AGTAGGAGG
+AGTAGGAGG

celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V3/bc2.txt

@@ -0,0 +1,96 @@
+TTGGTGACC
+CATCGGTTC
+CAATGCAAC
+AAGGTGGTA
+TCAGGTAGA
+CACTAGGCA
+GGCATGCAA
+ATGCGATAC
+AGACGAAGT
+CGATAAGGC
+CGAGTTGCA
+GGTGATCAG
+AGTGGTGGT
+TAGCGATGA
+ATGCCTAAG
+CCATAATCG
+CTAACCAGA
+TGATGTGCC
+TAGCTACAC
+TCCTGGCTT
+GTAGTTCCT
+GTATCCTTC
+GCCATAACC
+GCATGTTGG
+AATCCGGTG
+ACGTTACGA
+AGCATAGCG
+GGTCCGTAA
+GTTCTACCG
+CGCTGTAAG
+CGTCATACC
+TGGTAACCG
+ACAACAGGT
+TGACTTCCG
+CACGCAATA
+ACCGTACTC
+GATGTGTGT
+ACACCAACG
+TTGAGACAG
+CGGATCATC
+TAATGGCCG
+ACCTCGACT
+CGGCTAGAT
+TGGACTTGT
+TGATCCTCT
+ATAGCGTGT
+TGTCGGTGT
+AGCCACATA
+GAAGAAGCC
+CTGTGGTAT
+CAGCCGAAT
+ACCTGCTAC
+TCTTCTCAG
+GATCAGGAC
+TAGACCACT
+TGAGTAGTC
+TTCGAGGAT
+ATGTATCGG
+TCTGTCTGC
+GCTTACAGG
+ATACCAGTC
+AACGTCCAA
+CTCCTCAAT
+TTCCAATCG
+AACGCTAGT
+TTACACGAC
+CATACGACC
+CTAATCGCG
+CGTAATTGG
+GTAGTGTTG
+TATAGCGGT
+CGTACTGAA
+TCGATGTGG
+CCAGAAGAT
+AGGCTGTTG
+TGAGGCCTT
+CCAGTCCTA
+CAATTGCGC
+TTGCCGTCA
+GAGTTGACA
+TGGTCAGTT
+GGTAGTCCA
+ATCCTTCCA
+ATGTGCAGC
+GGAAGACTC
+GTAATGGAC
+TATCGTGCA
+CCTACAAGG
+TAGTCCGGA
+AGATACGCA
+ACTCATCGT
+GATAACCGC
+GCTGCGATA
+TACTCACCA
+ACCAGGTCA
+GCAACTTCA

celescope-2.2.2/celescope/data/chemistry/scopeV3.0.1/bclist → celescope-2.3.0.dev0/celescope/data/chemistry/GEXSCOPE-V3/bc3.txt

@@ -1,195 +1,3 @@
-AACGGACCT
-AGGACTCAC
-ACTGCCTAG
-GAACGCTAT
-GACTGGTTG
-CAGGACTTC
-TCGGTTCGT
-GTCTTGCGT
-TATCTCCGA
-CGCAACTAC
-CACTTCAGA
-GCTCTCACT
-ACGAAGCTC
-GTGTTAAGG
-GGCTCTCTA
-GCGTAGTAA
-AGCTCCTTG
-CACATTCAC
-TCCGTATCA
-CACCTGTAA
-ATGGTCTCA
-TGCATCAAG
-CGCCAATGA
-TCGACTGTC
-TGTGGACAC
-ACATCGGAC
-GCTTGAGGT
-GCCGTTATT
-AACACCGTT
-CCTAGTCTT
-AACACACAG
-CAACATCGG
-TACCTCTCC
-CCAATGTCT
-CGAGATAGT
-AGTTCAGAG
-ATCGACACG
-GACCTTAGC
-TAACCTACC
-TGGAGAACC
-GATGTTACG
-TGGCATGAG
-CTGGTACTT
-GCGAGTAAC
-GATCCATGC
-GGCTTCTCA
-CGTTAGCGT
-ACACAGGCT
-AACTGGCGA
-AGACGTTCA
-TGCGGTTCT
-CAGTCTTCG
-AGCTGAGTC
-GTGCATATC
-GATGGCTCA
-TGAGCGAAG
-CTTGACGTT
-GTTACTGGT
-GAGCAGCTT
-GTTGGAGTG
-TGCCTGATC
-GCAGATGTG
-CAGAGTACA
-TGTAGTGTG
-ACGAATGGA
-TCACTGGAA
-ATGACAGCA
-CTCAGAACT
-AAGCTTGCG
-TCGGACATG
-GAGGATTGA
-TCTTGGACA
-GTGCAAGGT
-CGATCGGTA
-GGATACCAC
-ATCGTTGGC
-TGGAACGTA
-GCCTACGAT
-CATGTAGGC
-CTCACGTTC
-TGTGAGTCA
-AATCGCCAC
-GTGCGACTA
-TCTGGCGAA
-ATACGCGGA
-CCTTGAATC
-CCACACATT
-CGGTGATTG
-AGGAGCAAT
-ACACACCAA
-GGATAGATG
-TACCGTCTG
-TCTTGCTTG
-CCAGCTAAC
-CTTACGCAG
-AGTAGGAGG
-TTGGTGACC
-CATCGGTTC
-CAATGCAAC
-AAGGTGGTA
-TCAGGTAGA
-CACTAGGCA
-GGCATGCAA
-ATGCGATAC
-AGACGAAGT
-CGATAAGGC
-CGAGTTGCA
-GGTGATCAG
-AGTGGTGGT
-TAGCGATGA
-ATGCCTAAG
-CCATAATCG
-CTAACCAGA
-TGATGTGCC
-TAGCTACAC
-TCCTGGCTT
-GTAGTTCCT
-GTATCCTTC
-GCCATAACC
-GCATGTTGG
-AATCCGGTG
-ACGTTACGA
-AGCATAGCG
-GGTCCGTAA
-GTTCTACCG
-CGCTGTAAG
-CGTCATACC
-TGGTAACCG
-ACAACAGGT
-TGACTTCCG
-CACGCAATA
-ACCGTACTC
-GATGTGTGT
-ACACCAACG
-TTGAGACAG
-CGGATCATC
-TAATGGCCG
-ACCTCGACT
-CGGCTAGAT
-TGGACTTGT
-TGATCCTCT
-ATAGCGTGT
-TGTCGGTGT
-AGCCACATA
-GAAGAAGCC
-CTGTGGTAT
-CAGCCGAAT
-ACCTGCTAC
-TCTTCTCAG
-GATCAGGAC
-TAGACCACT
-TGAGTAGTC
-TTCGAGGAT
-ATGTATCGG
-TCTGTCTGC
-GCTTACAGG
-ATACCAGTC
-AACGTCCAA
-CTCCTCAAT
-TTCCAATCG
-AACGCTAGT
-TTACACGAC
-CATACGACC
-CTAATCGCG
-CGTAATTGG
-GTAGTGTTG
-TATAGCGGT
-CGTACTGAA
-TCGATGTGG
-CCAGAAGAT
-AGGCTGTTG
-TGAGGCCTT
-CCAGTCCTA
-CAATTGCGC
-TTGCCGTCA
-GAGTTGACA
-TGGTCAGTT
-GGTAGTCCA
-ATCCTTCCA
-ATGTGCAGC
-GGAAGACTC
-GTAATGGAC
-TATCGTGCA
-CCTACAAGG
-TAGTCCGGA
-AGATACGCA
-ACTCATCGT
-GATAACCGC
-GCTGCGATA
-TACTCACCA
-ACCAGGTCA
-GCAACTTCA
 TTCGGTCAA
 GTACGGACT
 GGCCATGTT