updated cosmic version and fixed CDS mutations translation from COSMI…

…C, also fixed parameter shourtucs in ensembl download

pypgatk/cgenomes/cgenomes_proteindb.py

@@ -68,7 +68,7 @@ def get_multiple_options(options_str: str):
   def get_mut_pro_seq(snp, seq):
     nucleotide = ["A", "T", "C", "G"]
     mut_pro_seq = ""
-    if "?" not in snp.dna_mut:  # unambiguous DNA change known in CDS sequence
+    if "?" not in snp.dna_mut and snp.aa_mut!='p.?':  # unambiguous DNA change known in CDS sequence
       positions = re.findall(r'\d+', snp.dna_mut)
       if ">" in snp.dna_mut and len(positions) == 1:  # Substitution
         tmplist = snp.dna_mut.split(">")
@@ -149,19 +149,19 @@ def get_mut_pro_seq(snp, seq):
 
   def cosmic_to_proteindb(self):
     """
-        This function translate the mutation file + COSMIC genes into a protein Fasta database. The
-        method write into the file system the output Fasta.
+        This function translates the mutation file + COSMIC genes into a protein Fasta database. The
+        method writes into the file system the output Fasta.
         :return:
         """
-    self.get_logger().debug("Starting reading the All cosmic genes")
+    self.get_logger().debug("Starting reading All cosmic genes")
     COSMIC_CDS_DB = {}
     for record in SeqIO.parse(self._local_complete_genes, 'fasta'):
       try:
         COSMIC_CDS_DB[record.id].append(record)
       except KeyError:
         COSMIC_CDS_DB[record.id] = [record]
 
-    cosmic_input = open(self._local_mutation_file, encoding="latin-1")  # CosmicMutantExport.tsv
+    cosmic_input = open(self._local_mutation_file, encoding="latin-1")
 
     header = cosmic_input.readline().split("\t")
     regex = re.compile('[^a-zA-Z]')

pypgatk/commands/ensembl_downloader.py

@@ -25,10 +25,10 @@
 @click.option('-l', '--list_taxonomies',
               help='Print the list of all the taxonomies in ENSEMBL (https://www.ensembl.org)', is_flag=True,
               default=False)
-@click.option('-sg', '--skip_gtf', help="Skip the gtf file during the download", is_flag=True)
+@click.option('-sg', '--skip_gtf', help="Skip the GTF file during the download", is_flag=True)
 @click.option('-sp', '--skip_protein', help="Skip the protein fasta file during download", is_flag=True)
 @click.option('-sc', '--skip_cds', help='Skip the CDS file download', is_flag=True)
-@click.option('-sd', '--skip_cdna', help='Skip the cDNA file download', is_flag=True)
+@click.option('-sdn', '--skip_cdna', help='Skip the cDNA file download', is_flag=True)
 @click.option('-sn', '--skip_ncrna', help='Skip the ncRNA file download', is_flag=True)
 @click.option('-sd', '--skip_dna', help='Skip the DNA (reference genome assembly) file download', is_flag=True)
 @click.option('-sv', '--skip_vcf', help='Skip the VCF variant file', is_flag=True)

pypgatk/config/cosmic_config.yaml

@@ -2,10 +2,10 @@ cosmic_data:
   output_directory: database_cosmic
   cosmic_server:
     cosmic_ftp: https://cancer.sanger.ac.uk
-    mutations_url: cosmic/file_download/GRCh38/cosmic/v88
+    mutations_url: cosmic/file_download/GRCh38/cosmic/v94
     all_cds_genes_file: All_COSMIC_Genes.fasta.gz
     mutations_file: CosmicMutantExport.tsv.gz
-    mutations_cellline_url: cosmic/file_download/GRCh38/cell_lines/v92
+    mutations_cellline_url: cosmic/file_download/GRCh38/cell_lines/v94
     mutations_cellline_file: CosmicCLP_MutantExport.tsv.gz
     all_celllines_genes_file: All_CellLines_Genes.fasta.gz
     cosmic_user: ''