Merge pull request #65 from grexor/devel

Devel

pybio/__init__.py

@@ -112,7 +112,7 @@ def genome_prepare(species, genome_version, args):
     if pybio.utils.is_tool("STAR") and not args.nostar:
         star_folder = os.path.join(pybio.config.genomes_folder, f"{species}.assembly.{genome_version}.star")
         if not pybio.core.genomes.genomes_present.get(species, {}).get(genome_version, {}).get("STAR", False) or not os.path.exists(star_folder):
-            return_code = pybio.core.genomes.star_index(species, genome_version, threads=args.threads)
+            return_code = pybio.core.genomes.star_index(species, genome_version, args)
             if return_code==0:
                 pybio.core.genomes.genomes_present[species][genome_version]["STAR"] = True
                 # write gingo file

pybio/core/genomes.py

@@ -382,7 +382,7 @@ def download_annotation(species, genome_version):
     command = script.format(shell=pybio.config.shell, gtf_url=gtf_url, gdir=pybio.config.genomes_folder, species=species, genome_version=genome_version)
     return os.system(command)
 
-def star_index(species, genome_version, threads=1):
+def star_index(species, genome_version, args):
     species_capital = species.capitalize()
     assembly = species_db.get(species, {}).get("assembly", species)
     ensembl_version = genome_version.replace("ensembl", "")
@@ -394,12 +394,19 @@ def star_index(species, genome_version, threads=1):
 mkdir {species}.assembly.{genome_version}.star >/dev/null 2>&1
 cd {species}.assembly.{genome_version}.star
 gunzip -f -k ../{species}.annotation.{genome_version}/{species}.gtf.gz # -k to keep both .gz and uncompressed GTF, some tools require uncompressed GTF
-STAR --runMode genomeGenerate --genomeSAindexNbases {genomeSAindexNbases} --genomeDir ../{species}.assembly.{genome_version}.star --genomeFastaFiles ../{species}.assembly.{genome_version}/{species}.fasta --runThreadN {threads} --sjdbGTFfile ../{species}.annotation.{genome_version}/{species}.gtf
+STAR --runMode genomeGenerate {genomeChrBinNbits} --genomeSAindexNbases {genomeSAindexNbases} --genomeDir ../{species}.assembly.{genome_version}.star --genomeFastaFiles ../{species}.assembly.{genome_version}/{species}.fasta --runThreadN {threads} --sjdbGTFfile ../{species}.annotation.{genome_version}/{species}.gtf
 """
     fasta_file = f"{pybio.config.genomes_folder}/{species}.assembly.{genome_version}/{species}.fasta"
     fasta_size = os.path.getsize(fasta_file)
-    genomeSAindexNbases = int(min(14, math.log(fasta_size, 2)/2 - 1))
-    command = script.format(shell=pybio.config.shell, threads=threads, genomeSAindexNbases=genomeSAindexNbases, gdir=pybio.config.genomes_folder, species=species, species_capital=species_capital, assembly=assembly, ensembl_version=ensembl_version, genome_version=genome_version)
+    if not args.genomeSAindexNbases:
+        genomeSAindexNbases = int(min(14, math.log(fasta_size, 2)/2 - 1))
+    else:
+        genomeSAindexNbases = int(args.genomeSAindexNbases)
+    if args.genomeChrBinNbits:
+        genomeChrBinNbits = f"--genomeChrBinNbits {args.genomeChrBinNbits}"
+    else:
+        genomeChrBinNbits = ""
+    command = script.format(shell=pybio.config.shell, threads=args.threads, genomeChrBinNbits=genomeChrBinNbits, genomeSAindexNbases=genomeSAindexNbases, gdir=pybio.config.genomes_folder, species=species, species_capital=species_capital, assembly=assembly, ensembl_version=ensembl_version, genome_version=genome_version)
     return os.system(command)
 
 def salmon_index(species, genome_version):

pybio/pybio

@@ -17,6 +17,8 @@ parser.add_argument("-gtf", "--gtf", help="GTF file to import")
 parser.add_argument("-threads", "--threads", '-t', '--t', default="1", help="Number of threads to use (default: 1)")
 parser.add_argument("-help", "-h", "--help", action="store_true")
 parser.add_argument("-xs", "--xs", help="Add '--outSAMstrandField intronMotif'", action="store_false")
+parser.add_argument("-genomeSAindexNbases", "--genomeSAindexNbases", help="STAR genomeSAindexNbases")
+parser.add_argument("-genomeChrBinNbits", "--genomeChrBinNbits", help="STAR genomeChrBinNbits")
 args = parser.parse_args()
 
 help_0 = """
@@ -87,9 +89,9 @@ Example:
 $ pybio sam2bam file1.sam file1.bam
 """
 
-print(f"[pybio] v{pybio.version}, https://github.com/grexor/pybio")
-print(f"[pybio] config file: {pybio.config.config_fname()}")
-print(f"[pybio] genomes folder: {pybio.config.genomes_folder}")
+print(f"pybio | v{pybio.version}, https://github.com/grexor/pybio")
+print(f"pybio | config file: {pybio.config.config_fname()}")
+print(f"pybio | genomes folder: {pybio.config.genomes_folder}")
 print()
 
 if args.version:
@@ -118,12 +120,12 @@ def determine_species(species):
     return []
 
 def display_potential_hits(potential_hits, search_text):
-    print(f"We found {len(potential_hits)} genome hits for your provided genome species `{search_text}`.\n")
+    print(f"pybio found {len(potential_hits)} genome hits for your provided genome species `{search_text}`.\n")
     for hit in potential_hits:
-        print(f"Species = '{hit[1]}', display name = '{hit[2]}'")
+        print(f"Species = '\033[32;1m{hit[1]}\033[0m', display name = '{hit[2]}'")
     if len(potential_hits)>0:
         print(f"\nFor example, to download the first hit from the list above, you could write:\n")
-        print(f"$ pybio genome {potential_hits[0][1]}")
+        print(f"$ \033[32;1mpybio genome {potential_hits[0][1]}\033[0m")
         print()
     sys.exit(1)
 
@@ -141,7 +143,7 @@ def resolve_species_version(species, args, downloaded_only=False):
         genome_version = args.commands[2]
         return species.lower(), genome_version, potential_hits
     if len(args.commands)>=2:
-        if args.commands[0] not in ["search", "genome", "path", "species"]:
+        if args.commands[0] not in ["search", "genome", "genomes", "path", "species"]:
             genome_version = args.commands[1]
             return species.lower(), genome_version, potential_hits
     genome_version = determine_genome_version(species, downloaded_only=downloaded_only)
@@ -167,7 +169,7 @@ def determine_genome_version(species, downloaded_only):
     return pybio.core.genomes.species_db.get(species, {}).get("genome_version", None)
 
 def handle_genome(args):
-    search_string = args.commands[1] if args.commands[0] in ["genome", "search"] else args.commands[0]
+    search_string = args.commands[1] if args.commands[0] in ["genome", "genomes", "search"] else args.commands[0]
     species, genome_version, potential_hits = resolve_species_version(search_string, args)
     if len(potential_hits)>1:
         display_potential_hits(potential_hits, species)
@@ -225,6 +227,9 @@ if len(args.commands)>0:
         sys.exit()
 
     if args.commands[0] in ["genome", "genomes"]:
+        if len(args.commands)<2:
+            print("Please provide a search term for the species")
+            sys.exit()
         handle_genome(args)
         sys.exit()