def MoveGeneMarkFiles(workdir, genome):
"""
Handles temporary folders/files created by GeneMark-ES.
"""
# GeneMark-ES produces these filenames for each genome run.
to_move = ["data", "info", "output", "run", "gmes.log", "run.cfg",
"prot_seq.faa", "nuc_seq.fna", "genemark.gtf"]
# Attempt to make GeneMark-ES temporary file folder if not extant.
gmes = "{0}/gmes/{1}/".format(workdir, genome)
TryMkDirs(gmes)
# Move all files and folders to new folder.
logging.info("PanGuess: Moving/Removing GeneMark-ES temporary files and folders.")
for f in to_move:
if os.path.isdir(f):
if not os.path.isdir("{0}/{1}".format(gmes, f)):
shutil.move(f, gmes)
else:
shutil.rmtree(f)
elif os.path.isfile(f):
if not os.path.isfile("{0}/{1}".format(gmes, f)):
shutil.move(f, gmes)
else:
os.remove(f)
def ConstructGeneModelSets(attributes, exonerate_genes, workdir, genome, tag):
"""
Build completed gene model set for genome from our three sources.
"""
# Temporary gene/protein sets from GeneMark-ES and TransDecoder.
#gm_prot_db = SeqIO.index("{0}/gmes/{1}/prot_seq.faa".format(workdir, genome), "fasta")
#gm_nucl_db = SeqIO.index("{0}/gmes/{1}/nuc_seq.fna".format(workdir, genome), "fasta")
#td_prot_db = SeqIO.index("{0}/td/{1}/NCR.fna.transdecoder.pep".format(workdir, genome), "fasta")
#td_nucl_db = SeqIO.index("{0}/td/{1}/NCR.fna.transdecoder.cds".format(workdir, genome), "fasta")
# Master lists.
prot_models = []
nucl_models = []
# Try to make a directory for protein sets.
sdir = "{0}/sets".format(workdir)
TryMkDirs(sdir)
# Loop over attributes, extract gene from given source based on parent method.
for gene in attributes:
#if gene[4].startswith("TransDecoder"):
# prot_seq = td_prot_db[gene[1]]
# nucl_seq = td_nucl_db[gene[1]]
# prot_seq.id = "{0}|{1}_{2}_{3}".format(tag, gene[0], gene[2], gene[3])
# nucl_seq.id = prot_seq.id
# gene[1] = prot_seq.id
# prot_models.append(prot_seq)
# nucl_models.append(nucl_seq)
#elif gene[4].startswith("GeneMark"):
# prot_seq = gm_prot_db[gene[1]]
# nucl_seq = gm_nucl_db[gene[1]]
# prot_seq.id = "{0}|{1}_{2}_{3}".format(tag, gene[0], gene[2], gene[3])
# nucl_seq.id = prot_seq.id
# gene[1] = prot_seq.id
# prot_models.append(prot_seq)
# nucl_models.append(nucl_seq)
if gene[4].startswith("Exonerate"):
match = filter(lambda x: x.id == gene[1], exonerate_genes)
prot_seq = SeqRecord(Seq(match[0].prot), id=match[0].id)
nucl_seq = SeqRecord(Seq(match[0].nucl), id=match[0].id)
prot_seq.id = "{0}|{1}".format(tag, prot_seq.id)
nucl_seq.id = "{0}|{1}".format(tag, nucl_seq.id)
gene[1] = prot_seq.id
prot_models.append(prot_seq)
nucl_models.append(nucl_seq)
# Write protein sequences to file.
with open("{0}/{1}.faa".format(sdir, tag), "w") as outpro:
SeqIO.write(prot_models, outpro, "fasta")
# Write nucleotide sequences to file.
with open("{0}/{1}.nucl".format(sdir, tag), "w") as outnuc:
SeqIO.write(nucl_models, outnuc, "fasta")
# Write attributes to file.
with open("{0}/{1}.attributes".format(sdir, tag), "w") as outatt:
for line in attributes:
outatt.write("\t".join(str(el) for el in line) + "\n")
def BuildRefSet(workdir, ref):
"""
Build temporary set of reference proteins. It's faster to run Exonerate by splitting
up the dataset into individual files and running them as separate queries against
the genome than as a full file.
"""
# Make folder for reference proteins, if not already present.
ref_folder = "{0}/ref".format(workdir)
TryMkDirs(ref_folder)
# Split user-provided reference set into individual proteins (have to do this).
ref_db = SeqIO.index(ref, "fasta")
logging.info("PanGuess: Building reference protein sequence dataset.")
for seq in ref_db:
SeqIO.write(ref_db[seq], "{0}/{1}.faa".format(ref_folder, ref_db[seq].id), "fasta")
ref_db.close()
def KaryoPloteR(tags, karyotypes, lengths):
"""
Run Karyoplot.R for all strains in a dataset and write the plots to the karyplots folder.
"""
karyopath = os.path.dirname(os.path.realpath(sys.argv[0])) + "/Karyotype.R"
sp.call(["Rscript", karyopath, tags, karyotypes, lengths])
# Don't rewrite work directory if already there.
kdir = "./karyoplots"
TryMkDirs(kdir)
for tag in open(tags).readlines():
shutil.copy("{0}_components.eps".format(tag.strip("\n")), kdir)
shutil.copy("{0}_orthologs.eps".format(tag.strip("\n")), kdir)
os.remove("{0}_components.eps".format(tag.strip("\n")))
os.remove("{0}_orthologs.eps".format(tag.strip("\n")))
def RunBUSCO(buscopath, lineagepath, gene_sets):
"""
Runs BUSCO analysis on every protein set and writes output files to BUSCO folder.
"""
bdir = "./busco"
# Don't rewrite work directory if already there.
TryMkDirs(bdir)
for gene_set in gene_sets:
wd = gene_set.split("/")[-1]
cmd = [
buscopath, "-i", gene_set, "-l", lineagepath, "-o",
"{0}.busco".format(wd), "-m", "prot"
]
print "Running BUSCO"
sp.call(cmd)
shutil.move("run_{0}.busco".format(wd), bdir)
def RunTransDecoder(ncr, tp_path, tl_path, workdir, genome, td_len):
"""
Run the two TransDecoder commands via the command line.
"""
# Try to make a directory for TransDecoder. Might as well do it now.
tdir = "{0}/td/{1}/".format(workdir, genome)
TryMkDirs(tdir)
# Write NCRs to FASTA file
with open("{0}/NCR.fna".format(tdir), "w") as outfile:
for line in ncr:
outfile.write(line)
# Run both TransDecoder processes sequentially.
sp.call([tl_path, "-t", "{0}/NCR.fna".format(tdir), "-m", "{0}".format(td_len)])
sp.call([tp_path, "-t", "{0}/NCR.fna".format(tdir), "--single_best_only"])
# Return the TransDecoder directory for MoveTransDecoderFiles.
return tdir
def PanOCTOutputHandler():
"""
Move expected PanOCT output (might differ from what user actually specifies) to dedicated
PanOCT output directory.
"""
to_move = glob("*pairwise*") + glob("*cluster*") + glob("*paralog*") \
+ glob("*matchtable*") + ["centroids.fasta", "fragments_fusions.txt", "id.txt",
"missing_blast_results.txt", "parameters.txt", "report.txt"]
tdir = "panoct"
TryMkDirs(tdir)
for f in to_move:
if os.path.isdir(f):
if not os.path.isdir("{0}/{1}".format(tdir, f)):
shutil.move(f, tdir)
else:
shutil.rmtree(f)
elif os.path.isfile(f):
if not os.path.isfile("{0}/{1}".format(tdir, f)):
shutil.move(f, tdir)
else:
os.remove(f)
def MakeWorkingDirs():
"""
Tries to make work directory if not already present.
"""
tdir = "go"
TryMkDirs(tdir)
def MakeWorkingDir(workdir):
"""
Tries to make work directory if not already present.
"""
# Don't rewrite work directory if already there.
TryMkDirs(workdir)
def GenerateClusterFASTAs(genomes, refined=False):
"""
Extract gene model clusters from full database and write out nucleotide and protein sequence families to file.
"""
if not os.path.isfile("./gm_pred/sets/allnucl.db"):
ConcatenateDatasets(genomes)
elif not os.path.isfile("./gm_pred/sets/allprot.db"):
ConcatenateDatasets(genomes)
nt_index = SeqIO.index("./gm_pred/sets/allnucl.db", "fasta")
aa_index = SeqIO.index("./gm_pred/sets/allprot.db", "fasta")
fdir = "./panoct/clusters/"
matchtable = "./panoct/matchtable.txt"
TryMkDirs(fdir)
TryMkDirs("{0}/core/faa".format(fdir))
TryMkDirs("{0}/core/fna".format(fdir))
TryMkDirs("{0}/acc/faa".format(fdir))
TryMkDirs("{0}/acc/fna".format(fdir))
core, acc = ParseMatchtable(matchtable)
for cluster in core:
nt_seqs = [nt_index[member] for member in core[cluster]]
aa_seqs = [aa_index[member] for member in core[cluster]]
with open("{0}/core/fna/Core_{1}.fna".format(fdir, cluster),
"w") as aa_out:
SeqIO.write(nt_seqs, aa_out, "fasta")
with open("{0}/core/faa/Core_{1}.faa".format(fdir, cluster),
"w") as aa_out:
SeqIO.write(aa_seqs, aa_out, "fasta")
for cluster in acc:
nt_seqs = [nt_index[member] for member in acc[cluster] if member]
aa_seqs = [aa_index[member] for member in acc[cluster] if member]
with open("{0}/acc/fna/Acc_{1}.fna".format(fdir, cluster),
"w") as aa_out:
SeqIO.write(nt_seqs, aa_out, "fasta")
with open("{0}/acc/faa/Acc_{1}.faa".format(fdir, cluster),
"w") as aa_out:
SeqIO.write(aa_seqs, aa_out, "fasta")
if refined:
matchtable = "./panoct/refined_matchtable.txt"
rdir = "./panoct/clusters/refined"
TryMkDirs(rdir)
TryMkDirs("{0}/core/faa".format(rdir))
TryMkDirs("{0}/core/fna".format(rdir))
TryMkDirs("{0}/acc/faa".format(rdir))
TryMkDirs("{0}/acc/fna".format(rdir))
core, acc = ParseMatchtable(matchtable)
for cluster in core:
nt_seqs = [nt_index[member] for member in core[cluster]]
aa_seqs = [aa_index[member] for member in core[cluster]]
with open("{0}/core/fna/Core_{1}.fna".format(rdir, cluster),
"w") as aa_out:
SeqIO.write(nt_seqs, aa_out, "fasta")
with open("{0}/core/faa/Core_{1}.faa".format(rdir, cluster),
"w") as aa_out:
SeqIO.write(aa_seqs, aa_out, "fasta")
for cluster in acc:
nt_seqs = [nt_index[member] for member in acc[cluster] if member]
aa_seqs = [aa_index[member] for member in acc[cluster] if member]
with open("{0}/acc/fna/Acc_{1}.fna".format(rdir, cluster),
"w") as aa_out:
SeqIO.write(nt_seqs, aa_out, "fasta")
with open("{0}/acc/faa/Acc_{1}.faa".format(rdir, cluster),
"w") as aa_out:
SeqIO.write(aa_seqs, aa_out, "fasta")
def RemoveDubiousCalls(results, sets):
"""
"""
logging.info("QualityCheck: Filtering gene model sets for dubious calls.")
# Master list for calls to remove.
to_remove = []
# Loop through all QCBLAST results, flag top-hits that have >=70% sequence coverage with a dubious gene.
for result in results:
for query in result:
if query.hits:
query_len = query.seq_len
subj_len = query.hits[0].seq_len
ratio = min(query_len, subj_len) / max(query_len, subj_len)
if ratio >= 0.7:
to_remove.append(query.hits[0].id)
logging.info(
"QualityCheck: {0} has >=70% length overlap with {1}, assigning {0} as a"
" dubious call.".format(query.hits[0].id, query.id))
# Remove flagged calls from nucleotide and protein sets, and genomic attributes file.
for path in sets:
genome = path.split("/")[-1]
tag = genome.split(".")[0]
tr_strain = filter(lambda x: x.split("|")[0] == tag, to_remove)
if tr_strain:
aa_path = "./gm_pred/sets/{0}.faa".format(tag)
nt_path = "./gm_pred/sets/{0}.nucl".format(tag)
at_path = "./gm_pred/sets/{0}.attributes".format(tag)
current_prot = list(SeqIO.parse(open(aa_path), "fasta"))
current_nucl = list(SeqIO.parse(open(nt_path), "fasta"))
current_att = list(reader(open(at_path), delimiter="\t"))
to_move = [aa_path, nt_path, at_path]
TryMkDirs("./gm_pred/sets/old/")
new_prot = filter(lambda x: x.id not in tr_strain, current_prot)
new_nucl = filter(lambda x: x.id not in tr_strain, current_nucl)
new_att = filter(lambda x: x[1] not in tr_strain, current_att)
logging.info("QualityCheck: Removed {0} dubious calls from {1},"
" writing remaining calls to new files.".format(
len(tr_strain), genome))
logging.info("QualityCheck: Moving old calls.")
for f in to_move:
shutil.copy(f, "./gm_pred/sets/old/")
# Write protein sequences to file.
with open(aa_path, "w") as outpro:
SeqIO.write(new_prot, outpro, "fasta")
# Write nucleotide sequences to file.
with open(nt_path, "w") as outnuc:
SeqIO.write(new_nucl, outnuc, "fasta")
# Write attributes to file.
with open(at_path, "w") as outatt:
for line in new_att:
outatt.write("\t".join(str(el) for el in line) + "\n")
logging.info(
"QualityCheck: Completed removal of dubious calls from all datasets.")
