def dnld_refseqs_for_taxid(taxid,
filter_term,
taxonomy,
dir_cache_refseqs,
query='',
db='nuccore'):
ft = None
if filter_term == 'plastid':
ft = '("chloroplast"[filter] OR "plastid"[filter])'
else:
ft = '("' + filter_term + '"[filter])'
tax_terms = tuple(reversed(taxonomy.lineage_for_taxid(taxid)['names']))
for tax_term in tax_terms:
if tax_term is None:
tax_term = taxonomy.scientific_name_for_taxid(taxid)
term = '"RefSeq"[Keyword] AND "{}"[Primary Organism] AND {}'.format(
tax_term, ft)
term = query + term
accs = set(accs_eutil(search_eutil(db, term)))
if len(accs) > 0:
plural = 'sequences'
if len(accs) == 1:
plural = 'sequence'
Log.msg(
'Found {} RefSeq {} {} for'.format(len(accs), filter_term,
plural), tax_term)
# Random sample ###################################################
if len(accs) > 10:
Log.wrn('Using a random sample of ten RefSeq sequences.')
random.seed(a=len(accs), version=2)
accs = set(random.sample(accs, 10))
###################################################################
break
else:
Log.wrn(
'No RefSeq {} sequences were found for'.format(filter_term),
tax_term)
cache_path = opj(
dir_cache_refseqs,
filter_term + '__' + tax_term.replace(' ', '_') + '.fasta')
parsed_fasta_cache = {}
if ope(cache_path):
parsed_fasta_cache = read_fasta(cache_path,
seq_type=SEQ_TYPE_NT,
def_to_first_space=True)
parsed_fasta_cache = seq_records_to_dict(parsed_fasta_cache)
for acc in parsed_fasta_cache:
if acc in accs:
accs.remove(acc)
if len(accs) > 0:
parsed_fasta = dnld_ncbi_seqs(db, list(accs))
parsed_fasta = seq_records_to_dict(parsed_fasta, prepend_acc=True)
parsed_fasta.update(parsed_fasta_cache)
write_fasta(parsed_fasta, cache_path)
return cache_path
def _process_downloaded_seq_data(efetch_txt: str, db: str, rettype: str,
retmode: str):
rec_list = list()
if rettype == 'gb' and retmode == 'xml':
rec_list = seq_records_gb(efetch_txt)
elif rettype == 'fasta' and retmode == 'text':
seq_type = None
if db == 'nuccore':
seq_type = SEQ_TYPE_NT
elif db == 'protein':
seq_type = SEQ_TYPE_AA
rec_list = read_fasta(StringIO(efetch_txt), seq_type, parse_def=True)
return rec_list
def dnld_prot_seqs(ss, prot_acc_user, aa_prot_ncbi_file, dir_cache_prj):
if len(prot_acc_user) != 0:
acc_old = set()
if ope(aa_prot_ncbi_file):
_ = read_fasta(aa_prot_ncbi_file, SEQ_TYPE_AA)
acc_old = set([x.definition.split('|')[0] for x in _])
if acc_old == set(prot_acc_user):
return prot_acc_user
else:
pickle_file = opj(dir_cache_prj, 'ncbi_prot_metadata_cache__' + ss)
if ope(pickle_file):
with open(pickle_file, 'rb') as f:
pa_info = pickle.load(f)
print()
Log.inf('Downloading protein sequences from NCBI:', ss)
_ = dnld_ncbi_seqs('protein',
prot_acc_user,
rettype='gb',
retmode='xml')
prot_acc_user_new = list()
for rec in _:
acc_ver = rec.accession_version
defn = rec.definition
organism = rec.organism
prot_acc_user_new.append(acc_ver)
defn_new = defn.split('[' + organism + ']')[0]
defn_new = defn_new.lower().strip()
defn_new = defn_new.replace(' ', '_').replace('-', '_')
defn_new = defn_new.replace(',', '')
defn_new = defn_new[0].upper() + defn_new[1:]
defn_new = acc_ver + '|' + defn_new + '|' + organism
defn_new = defn_new.replace(' ', '_').replace('-', '_')
rec.definition = defn_new
prot_acc_user = prot_acc_user_new
write_fasta(_, aa_prot_ncbi_file)
else:
if ope(aa_prot_ncbi_file):
osremove(aa_prot_ncbi_file)
return prot_acc_user
def filter_queries(ss,
aa_queries_file,
min_query_length,
max_query_length,
max_query_identity,
vsearch,
prot_acc_user,
overwrite,
logging=True):
if logging is True:
print()
Log.inf('Filtering AA query sequences:', ss)
Log.msg('min_query_length:', str(min_query_length))
Log.msg('max_query_length:', str(max_query_length))
Log.msg('max_query_identity:', str(max_query_identity))
parsed_fasta_1 = filter_fasta_by_length(aa_queries_file, SEQ_TYPE_AA,
min_query_length, max_query_length)
tmp1 = aa_queries_file + '_temp1'
tmp2 = aa_queries_file + '_temp2'
for rec in parsed_fasta_1:
rec.seq.gc_code = 1
rec.seq = rec.seq.untranslate()
write_fasta(parsed_fasta_1, tmp1)
run_cluster_fast(vsearch, max_query_identity, tmp1, tmp2)
parsed_fasta_2 = read_fasta(tmp2, SEQ_TYPE_DNA, parse_def=True)
prot_acc_user_new = list()
for rec in parsed_fasta_2:
rec.seq.gc_code = 1
rec.seq = rec.seq.translate()
acc = rec.accession_version
if acc in prot_acc_user:
prot_acc_user_new.append(acc)
if overwrite is True:
write_fasta(parsed_fasta_2, aa_queries_file, prepend_acc=True)
osremove(tmp1)
osremove(tmp2)
return prot_acc_user_new
def run_tblastn_on_assemblies(ss, assemblies, aa_queries_file, tblastn,
dir_prj_assmbl_blast_results, blast_2_evalue,
blast_2_max_hsps, blast_2_qcov_hsp_perc,
blast_2_best_hit_overhang,
blast_2_best_hit_score_edge,
blast_2_max_target_seqs, threads, dir_cache_prj,
dir_prj_ips):
if len(assemblies) > 0:
print()
Log.inf('Running BLAST on assemblies:', ss)
if tblastn is None:
Log.err('tblastn is not available. Cannot continue. Exiting.')
exit(0)
else:
Log.wrn('There are no assemblies. Nothing to do, stopping.')
exit(0)
cache_file = opj(dir_cache_prj, 'blast_2_settings_cache__' + ss)
pickled = dict()
settings = {'blast_2_evalue': blast_2_evalue,
'blast_2_max_hsps': blast_2_max_hsps,
'blast_2_qcov_hsp_perc': blast_2_qcov_hsp_perc,
'blast_2_best_hit_overhang': blast_2_best_hit_overhang,
'blast_2_best_hit_score_edge': blast_2_best_hit_score_edge,
'blast_2_max_target_seqs': blast_2_max_target_seqs,
'queries': seq_records_to_dict(
read_fasta(aa_queries_file, SEQ_TYPE_AA))}
Log.msg('evalue:', str(blast_2_evalue))
Log.msg('max_hsps:', str(blast_2_max_hsps))
Log.msg('qcov_hsp_perc:', str(blast_2_qcov_hsp_perc))
Log.msg('best_hit_overhang:', str(blast_2_best_hit_overhang))
Log.msg('best_hit_score_edge:', str(blast_2_best_hit_score_edge))
Log.msg('max_target_seqs:', str(blast_2_max_target_seqs))
print()
for a in assemblies:
assmbl_src = a['src']
assmbl_name = a['name']
if assmbl_src != 'user_fasta':
if assmbl_name.endswith('__' + ss):
assmbl_name = assmbl_name.replace('__' + ss, '')
else:
continue
assmbl_blast_db_path = a['blast_db_path']
assmbl_genetic_code = a['gc_id']
ips_json_dump_path = opj(dir_prj_ips, assmbl_name + '_ann_ips__' + ss +
'.json')
_ = opj(dir_prj_assmbl_blast_results, assmbl_name + '__' + ss + '.tsv')
if ope(_) and ope(cache_file):
with open(cache_file, 'rb') as f:
pickled = pickle.load(f)
if ope(_) and pickled == settings:
# Log.msg('The provided BLAST settings and query sequences did '
# 'not change since the previous run.')
Log.msg('BLAST results already exist:', assmbl_name)
else:
Log.msg('Running tblastn on: ' + assmbl_name, ss)
if ope(ips_json_dump_path):
osremove(ips_json_dump_path)
run_blast(exec_file=tblastn,
task='tblastn',
threads=threads,
db_path=assmbl_blast_db_path,
queries_file=aa_queries_file,
out_file=_,
evalue=blast_2_evalue,
max_hsps=blast_2_max_hsps,
qcov_hsp_perc=blast_2_qcov_hsp_perc,
best_hit_overhang=blast_2_best_hit_overhang,
best_hit_score_edge=blast_2_best_hit_score_edge,
max_target_seqs=blast_2_max_target_seqs,
db_genetic_code=assmbl_genetic_code,
out_cols=BLST_RES_COLS_2)
a['blast_hits_aa__' + ss] = parse_blast_results_file(_, BLST_RES_COLS_2)
with open(cache_file, 'wb') as f:
pickle.dump(settings, f, protocol=PICKLE_PROTOCOL)
def main():
"""Run the script."""
# Prepare initial logger (before we know the log file path) --------------
prj_log_file_suffix = time_stamp() + '.log'
log_stream = StringIO()
Log.set_colors(COLORS)
Log.set_file(log_stream)
Log.set_write(True)
# Prepare configuration directory ----------------------------------------
if ope(DIR_CFG):
Log.inf('Found configuration directory:', DIR_CFG)
else:
Log.wrn('Creating configuration directory:', DIR_CFG)
make_dirs(DIR_CFG)
print()
# Check for dependencies -------------------------------------------------
Log.inf('Checking for dependencies.')
make_dirs(DIR_DEP)
make_dirs(DIR_KRK)
seqtk = deps.dep_check_seqtk(DIR_DEP, FORCE_DEPS)
trimmomatic, adapters = deps.dep_check_trimmomatic(DIR_DEP)
fasterq_dump = deps.dep_check_sra_toolkit(DIR_DEP, OS_ID, DIST_ID,
DEBIAN_DISTS, REDHAT_DISTS,
FORCE_DEPS)
makeblastdb, _, tblastn = deps.dep_check_blast(DIR_DEP, OS_ID, DIST_ID,
DEBIAN_DISTS, REDHAT_DISTS,
FORCE_DEPS)
vsearch = deps.dep_check_vsearch(DIR_DEP, OS_ID, DIST_ID, DEBIAN_DISTS,
REDHAT_DISTS, FORCE_DEPS)
spades = deps.dep_check_spades(DIR_DEP, OS_ID, FORCE_DEPS)
bowtie2, bowtie2_build = deps.dep_check_bowtie2(DIR_DEP, OS_ID, FORCE_DEPS)
rcorrector = deps.dep_check_rcorrector(DIR_DEP, FORCE_DEPS)
kraken2, kraken2_build = deps.dep_check_kraken2(DIR_DEP, OS_ID,
RELEASE_NAME, FORCE_DEPS)
print()
kraken2_dbs = deps.dnld_kraken2_dbs(DIR_KRK)
if INSTALL_DEPS is True or DNLD_KRAKEN_DBS is True:
exit(0)
print()
# Initialize NCBI taxonomy database --------------------------------------
tax = Taxonomy()
if tax.is_initialized() is False:
tax.init(data_dir_path=DIR_TAX, logger=Log)
print()
# Parse configuration file -----------------------------------------------
Log.inf('Reading configuration file:', CONFIG_FILE_PATH)
_ = config_file_parse(CONFIG_FILE_PATH, tax)
allow_no_stop_cod = _['allow_no_stop_cod']
allow_no_strt_cod = _['allow_no_strt_cod']
allow_non_aug = _['allow_non_aug']
blast_1_evalue = _['blast_1_evalue']
blast_1_max_hsps = _['blast_1_max_hsps']
blast_1_qcov_hsp_perc = _['blast_1_qcov_hsp_perc']
blast_1_best_hit_overhang = _['blast_1_best_hit_overhang']
blast_1_best_hit_score_edge = _['blast_1_best_hit_score_edge']
blast_1_max_target_seqs = _['blast_1_max_target_seqs']
blast_2_evalue = _['blast_2_evalue']
blast_2_max_hsps = _['blast_2_max_hsps']
blast_2_qcov_hsp_perc = _['blast_2_qcov_hsp_perc']
blast_2_best_hit_overhang = _['blast_2_best_hit_overhang']
blast_2_best_hit_score_edge = _['blast_2_best_hit_score_edge']
blast_2_max_target_seqs = _['blast_2_max_target_seqs']
dir_out = _['output_directory']
email = _['email']
requery_after = _['requery_after']
fq_pe = _['fq_pe']
fq_se = _['fq_se']
should_run_rcorrector = _['should_run_rcorrector']
should_run_ipr = _['should_run_ipr']
bt2_order = _['bt2_order']
kraken_confidence = _['kraken_confidence']
krkn_order = _['krkn_order']
prepend_assmbl = _['prepend_assmbl']
prj_name = _['project_name']
sras = _['sras']
tax_group = _['tax_group']
# tax_group_name = _['tax_group_name']
tax_ids_user = _['tax_ids']
user_assemblies = _['assmbl']
print()
# Parse search strategies file -------------------------------------------
if SS_FILE_PATH is not None:
Log.inf('Reading search strategies file:', SS_FILE_PATH)
sss = ss_file_parse(SS_FILE_PATH)
else:
Log.wrn('Search strategies file was not provided.\n' +
'Will process reads, assemblies and then stop.')
sss = dict()
print()
# Create output directory ------------------------------------------------
if dir_out is not None:
if ope(dir_out):
Log.inf('Found output directory:', dir_out)
else:
Log.wrn('Creating output directory:', dir_out)
make_dirs(dir_out)
print()
# Write Kakapo version information to the output directory ---------------
version_file = opj(dir_out, 'kakapo_version.txt')
if ope(version_file):
with open(version_file, 'r') as f:
version_prev = f.read().strip()
if __version__ != version_prev:
Log.wrn('The output directory contains data produced by a ' +
'different version of Kakapo: ' + version_prev +
'.\nThe currently running version is: ' + __version__ +
'.\n' +
'Delete "kakapo_version.txt" file located in the ' +
'output directory if you would like to continue.')
exit(0)
with open(version_file, 'w') as f:
f.write(__version__)
# Create subdirectories in the output directory --------------------------
_ = prepare_output_directories(dir_out, prj_name)
dir_temp = _['dir_temp']
dir_cache_pfam_acc = _['dir_cache_pfam_acc']
dir_cache_fq_minlen = _['dir_cache_fq_minlen']
dir_cache_prj = _['dir_cache_prj']
dir_cache_refseqs = _['dir_cache_refseqs']
dir_prj_logs = _['dir_prj_logs']
dir_prj_queries = _['dir_prj_queries']
dir_fq_data = _['dir_fq_data']
dir_fq_cor_data = _['dir_fq_cor_data']
dir_fq_trim_data = _['dir_fq_trim_data']
dir_fq_filter_bt2_data = _['dir_fq_filter_bt2_data']
dir_fq_filter_krkn2_data = _['dir_fq_filter_krkn2_data']
dir_fa_trim_data = _['dir_fa_trim_data']
dir_blast_fa_trim = _['dir_blast_fa_trim']
dir_prj_blast_results_fa_trim = _['dir_prj_blast_results_fa_trim']
dir_prj_vsearch_results_fa_trim = _['dir_prj_vsearch_results_fa_trim']
dir_prj_spades_assemblies = _['dir_prj_spades_assemblies']
dir_prj_blast_assmbl = _['dir_prj_blast_assmbl']
dir_prj_assmbl_blast_results = _['dir_prj_assmbl_blast_results']
dir_prj_transcripts = _['dir_prj_transcripts']
dir_prj_ips = _['dir_prj_ips']
dir_prj_transcripts_combined = _['dir_prj_transcripts_combined']
# Prepare logger ---------------------------------------------------------
prj_log_file = opj(dir_prj_logs, prj_name + '_' + prj_log_file_suffix)
with open(prj_log_file, 'w') as f:
f.write(SCRIPT_INFO.strip() + '\n\n' + log_stream.getvalue())
Log.set_colors(COLORS)
Log.set_file(prj_log_file)
Log.set_write(True)
log_stream.close()
# Resolve descending taxonomy nodes --------------------------------------
tax_ids = tax.all_descending_taxids_for_taxids([tax_group])
# Pfam uniprot accessions ------------------------------------------------
pfam_uniprot_acc = OrderedDict()
for ss in sss:
pfam_acc = sss[ss]['pfam_families']
pfam_uniprot_acc[ss] = pfam_uniprot_accessions(ss, pfam_acc, tax_ids,
dir_cache_pfam_acc)
# Download Pfam uniprot sequences if needed ------------------------------
aa_uniprot_files = OrderedDict()
for ss in sss:
aa_uniprot_files[ss] = opj(dir_prj_queries,
'aa_uniprot__' + ss + '.fasta')
# ToDo: add support for the requery_after parameter.
dnld_pfam_uniprot_seqs(ss, pfam_uniprot_acc[ss], aa_uniprot_files[ss],
dir_cache_prj)
# User provided entrez query ---------------------------------------------
prot_acc_user_from_query = OrderedDict()
for ss in sss:
entrez_queries = sss[ss]['entrez_search_queries']
prot_acc_user_from_query[ss] = user_entrez_search(
ss, entrez_queries, dir_cache_prj, requery_after)
# User provided protein accessions ---------------------------------------
prot_acc_user = OrderedDict()
for ss in sss:
print()
prot_acc_all = sorted(
set(sss[ss]['ncbi_accessions_aa'] + prot_acc_user_from_query[ss]))
prot_acc_user[ss] = user_protein_accessions(ss, prot_acc_all,
dir_cache_prj, tax)
# Download from NCBI if needed -------------------------------------------
aa_prot_ncbi_files = OrderedDict()
for ss in sss:
aa_prot_ncbi_files[ss] = opj(dir_prj_queries,
'aa_prot_ncbi__' + ss + '.fasta')
prot_acc_user[ss] = dnld_prot_seqs(ss, prot_acc_user[ss],
aa_prot_ncbi_files[ss],
dir_cache_prj)
# User provided protein sequences ----------------------------------------
aa_prot_user_files = OrderedDict()
for ss in sss:
user_queries = sss[ss]['fasta_files_aa']
aa_prot_user_files[ss] = opj(dir_prj_queries,
'aa_prot_user__' + ss + '.fasta')
user_aa_fasta(ss, user_queries, aa_prot_user_files[ss])
# Combine all AA queries -------------------------------------------------
print()
aa_queries_files = OrderedDict()
for ss in sss:
aa_queries_files[ss] = opj(dir_prj_queries, 'aa_all__' + ss + '.fasta')
combine_aa_fasta(ss, [
aa_uniprot_files[ss], aa_prot_ncbi_files[ss],
aa_prot_user_files[ss]
], aa_queries_files[ss])
# Filter AA queries ------------------------------------------------------
prot_acc_user_filtered = OrderedDict()
for ss in sss:
min_query_length = sss[ss]['min_query_length']
max_query_length = sss[ss]['max_query_length']
max_query_identity = sss[ss]['max_query_identity']
# Dereplicate all queries
filter_queries(ss,
aa_queries_files[ss],
min_query_length,
max_query_length,
max_query_identity,
vsearch,
prot_acc_user[ss],
overwrite=True)
# Dereplicate only NCBI queries. CDS for these will be downloaded
# later for reference.
if ope(aa_prot_ncbi_files[ss]):
prot_acc_user_filtered[ss] = filter_queries(ss,
aa_prot_ncbi_files[ss],
min_query_length,
max_query_length,
max_query_identity,
vsearch,
prot_acc_user[ss],
overwrite=False,
logging=False)
# Download SRA run metadata if needed ------------------------------------
sra_runs_info, sras_acceptable = dnld_sra_info(sras, dir_cache_prj)
# Download SRA run FASTQ files if needed ---------------------------------
x, y, z = dnld_sra_fastq_files(sras_acceptable, sra_runs_info, dir_fq_data,
fasterq_dump, THREADS, dir_temp)
se_fastq_files_sra = x
pe_fastq_files_sra = y
sra_runs_info = z
# User provided FASTQ files ----------------------------------------------
se_fastq_files_usr, pe_fastq_files_usr = user_fastq_files(fq_se, fq_pe)
# Collate FASTQ file info ------------------------------------------------
se_fastq_files = se_fastq_files_sra.copy()
se_fastq_files.update(se_fastq_files_usr)
pe_fastq_files = pe_fastq_files_sra.copy()
pe_fastq_files.update(pe_fastq_files_usr)
def gc_tt(k, d, tax):
taxid = d[k]['tax_id']
gc = tax.genetic_code_for_taxid(taxid)
d[k]['gc_id'] = gc
d[k]['gc_tt'] = TranslationTable(gc)
gc_mito = None
tt_mito = None
gc_plastid = None
tt_plastid = None
if tax.is_eukaryote(taxid) is True:
gc_mito = tax.mito_genetic_code_for_taxid(taxid)
if gc_mito != '0':
tt_mito = TranslationTable(gc_mito)
if tax.contains_plastid(taxid) is True:
gc_plastid = tax.plastid_genetic_code_for_taxid(taxid)
if gc_plastid != '0':
tt_plastid = TranslationTable(gc_plastid)
d[k]['gc_id_mito'] = gc_mito
d[k]['gc_tt_mito'] = tt_mito
d[k]['gc_id_plastid'] = gc_plastid
d[k]['gc_tt_plastid'] = tt_plastid
for se in se_fastq_files:
gc_tt(se, se_fastq_files, tax)
for pe in pe_fastq_files:
gc_tt(pe, pe_fastq_files, tax)
# Minimum acceptable read length -----------------------------------------
min_accept_read_len(se_fastq_files, pe_fastq_files, dir_temp,
dir_cache_fq_minlen, vsearch)
# Run Rcorrector ---------------------------------------------------------
run_rcorrector(se_fastq_files, pe_fastq_files, dir_fq_cor_data, rcorrector,
THREADS, dir_temp, should_run_rcorrector)
# File name patterns -----------------------------------------------------
a, b, c, d, e = file_name_patterns()
pe_trim_fq_file_patterns = a
pe_trim_fa_file_patterns = b
pe_blast_db_file_patterns = c
pe_blast_results_file_patterns = d
pe_vsearch_results_file_patterns = e
# Run Trimmomatic --------------------------------------------------------
run_trimmomatic(se_fastq_files, pe_fastq_files, dir_fq_trim_data,
trimmomatic, adapters, pe_trim_fq_file_patterns, THREADS)
# Run Bowtie 2 -----------------------------------------------------------
run_bt2_fq(se_fastq_files, pe_fastq_files, dir_fq_filter_bt2_data, bowtie2,
bowtie2_build, THREADS, dir_temp, bt2_order,
pe_trim_fq_file_patterns, tax, dir_cache_refseqs)
# Run Kraken2 ------------------------------------------------------------
run_kraken2(krkn_order, kraken2_dbs, se_fastq_files, pe_fastq_files,
dir_fq_filter_krkn2_data, kraken_confidence, kraken2, THREADS,
dir_temp, pe_trim_fq_file_patterns)
se_fastq_files = OrderedDict(se_fastq_files)
pe_fastq_files = OrderedDict(pe_fastq_files)
se_fastq_files = OrderedDict(
sorted(se_fastq_files.items(), key=lambda x: x[1]['filter_path_fq']))
pe_fastq_files = OrderedDict(
sorted(pe_fastq_files.items(), key=lambda x: x[1]['filter_path_fq']))
# Stop After Filter ------------------------------------------------------
if STOP_AFTER_FILTER is True:
Log.wrn('Stopping after Kraken2/Bowtie2 filtering step as requested.')
exit(0)
# Convert filtered FASTQ files to FASTA ----------------------------------
filtered_fq_to_fa(se_fastq_files, pe_fastq_files, dir_fa_trim_data, seqtk,
pe_trim_fa_file_patterns)
# Run makeblastdb on reads -----------------------------------------------
makeblastdb_fq(se_fastq_files, pe_fastq_files, dir_blast_fa_trim,
makeblastdb, pe_blast_db_file_patterns)
# Check if there are any query sequences.
any_queries = False
for ss in sss:
if stat(aa_queries_files[ss]).st_size == 0:
continue
else:
any_queries = True
# Run tblastn on reads ---------------------------------------------------
for ss in sss:
if stat(aa_queries_files[ss]).st_size == 0:
continue
changed_blast_1 = run_tblastn_on_reads(
se_fastq_files, pe_fastq_files, aa_queries_files[ss], tblastn,
blast_1_evalue, blast_1_max_hsps, blast_1_qcov_hsp_perc,
blast_1_best_hit_overhang, blast_1_best_hit_score_edge,
blast_1_max_target_seqs, dir_prj_blast_results_fa_trim,
pe_blast_results_file_patterns, ss, THREADS, seqtk, vsearch,
dir_cache_prj)
if changed_blast_1 is True:
if ope(dir_prj_vsearch_results_fa_trim):
rmtree(dir_prj_vsearch_results_fa_trim)
if ope(dir_prj_spades_assemblies):
rmtree(dir_prj_spades_assemblies)
if ope(dir_prj_blast_assmbl):
rmtree(dir_prj_blast_assmbl)
if ope(dir_prj_assmbl_blast_results):
rmtree(dir_prj_assmbl_blast_results)
if ope(dir_prj_transcripts):
rmtree(dir_prj_transcripts)
if ope(dir_prj_transcripts_combined):
rmtree(dir_prj_transcripts_combined)
prepare_output_directories(dir_out, prj_name)
# Run vsearch on reads ---------------------------------------------------
# should_run_vsearch = False
# for ss in sss:
# if stat(aa_queries_files[ss]).st_size == 0:
# continue
# else:
# should_run_vsearch = True
# break
# if should_run_vsearch is True:
# print()
# Log.inf('Checking if Vsearch should be run.')
for ss in sss:
if stat(aa_queries_files[ss]).st_size == 0:
continue
print()
Log.inf('Checking if Vsearch should be run:', ss)
run_vsearch_on_reads(se_fastq_files, pe_fastq_files, vsearch,
dir_prj_vsearch_results_fa_trim,
pe_vsearch_results_file_patterns, ss, seqtk)
# Run SPAdes -------------------------------------------------------------
# should_run_spades = False
# for ss in sss:
# if stat(aa_queries_files[ss]).st_size == 0:
# continue
# else:
# should_run_spades = True
# break
# if should_run_spades is True:
# print()
# Log.inf('Checking if SPAdes should be run.')
for ss in sss:
if stat(aa_queries_files[ss]).st_size == 0:
for se in se_fastq_files:
se_fastq_files[se]['spades_assembly' + '__' + ss] = None
for pe in pe_fastq_files:
pe_fastq_files[pe]['spades_assembly' + '__' + ss] = None
continue
print()
Log.inf('Checking if SPAdes should be run:', ss)
run_spades(se_fastq_files, pe_fastq_files, dir_prj_spades_assemblies,
spades, dir_temp, ss, THREADS, RAM)
# Combine SPAdes and user provided assemblies ----------------------------
assemblies = combine_assemblies(se_fastq_files, pe_fastq_files,
user_assemblies, tax, sss)
# Run makeblastdb on assemblies -----------------------------------------
makeblastdb_assemblies(assemblies, dir_prj_blast_assmbl, makeblastdb)
if any_queries is False:
Log.wrn('No query sequences were provided.')
# Run tblastn on assemblies ----------------------------------------------
for ss in sss:
if stat(aa_queries_files[ss]).st_size == 0:
continue
should_run_tblastn = False
for a in assemblies:
assmbl_src = a['src']
assmbl_name = a['name']
if assmbl_src != 'user_fasta':
if assmbl_name.endswith('__' + ss):
should_run_tblastn = True
break
else:
should_run_tblastn = True
break
if should_run_tblastn is False:
print()
Log.inf('Will not run BLAST. No transcripts exist:', ss)
continue
blast_2_evalue_ss = sss[ss]['blast_2_evalue']
blast_2_max_hsps_ss = sss[ss]['blast_2_max_hsps']
blast_2_qcov_hsp_perc_ss = sss[ss]['blast_2_qcov_hsp_perc']
blast_2_best_hit_overhang_ss = sss[ss]['blast_2_best_hit_overhang']
blast_2_best_hit_score_edge_ss = sss[ss]['blast_2_best_hit_score_edge']
blast_2_max_target_seqs_ss = sss[ss]['blast_2_max_target_seqs']
if blast_2_evalue_ss is None:
blast_2_evalue_ss = blast_2_evalue
if blast_2_max_hsps_ss is None:
blast_2_max_hsps_ss = blast_2_max_hsps
if blast_2_qcov_hsp_perc_ss is None:
blast_2_qcov_hsp_perc_ss = blast_2_qcov_hsp_perc
if blast_2_best_hit_overhang_ss is None:
blast_2_best_hit_overhang_ss = blast_2_best_hit_overhang
if blast_2_best_hit_score_edge_ss is None:
blast_2_best_hit_score_edge_ss = blast_2_best_hit_score_edge
if blast_2_max_target_seqs_ss is None:
blast_2_max_target_seqs_ss = blast_2_max_target_seqs
run_tblastn_on_assemblies(
ss, assemblies, aa_queries_files[ss], tblastn,
dir_prj_assmbl_blast_results, blast_2_evalue_ss,
blast_2_max_hsps_ss, blast_2_qcov_hsp_perc_ss,
blast_2_best_hit_overhang_ss, blast_2_best_hit_score_edge_ss,
blast_2_max_target_seqs_ss, THREADS, dir_cache_prj, dir_prj_ips)
# Prepare BLAST hits for analysis: find ORFs, translate ------------------
for ss in sss:
if stat(aa_queries_files[ss]).st_size == 0:
continue
min_target_orf_len_ss = sss[ss]['min_target_orf_length']
max_target_orf_len_ss = sss[ss]['max_target_orf_length']
organelle = sss[ss]['organelle']
blast_2_qcov_hsp_perc_ss = sss[ss]['blast_2_qcov_hsp_perc']
if blast_2_qcov_hsp_perc_ss is None:
blast_2_qcov_hsp_perc_ss = blast_2_qcov_hsp_perc
find_orfs_translate(ss, assemblies, dir_prj_transcripts, seqtk,
dir_temp, prepend_assmbl, min_target_orf_len_ss,
max_target_orf_len_ss, allow_non_aug,
allow_no_strt_cod, allow_no_stop_cod, tax,
tax_group, tax_ids_user, blast_2_qcov_hsp_perc_ss,
organelle)
# GFF3 files from kakapo results JSON files ------------------------------
# print()
for ss in sss:
if stat(aa_queries_files[ss]).st_size == 0:
continue
gff_from_json(ss, assemblies, dir_prj_ips,
dir_prj_transcripts_combined, prj_name)
# Run InterProScan 5 -----------------------------------------------------
if should_run_ipr is True:
print()
ss_names = tuple(sss.keys())
# Determine the length of printed strings, for better spacing --------
max_title_a_len = 0
max_run_id_len = 0
for a in assemblies:
for ss in ss_names:
if 'transcripts_aa_orf_fasta_file__' + ss not in a:
continue
aa_file = a['transcripts_aa_orf_fasta_file__' + ss]
if aa_file is None:
continue
assmbl_name = a['name']
run_id = ss + '_' + assmbl_name
max_run_id_len = max(len(run_id), max_run_id_len)
seqs = seq_records_to_dict(read_fasta(aa_file, SEQ_TYPE_AA))
# Filter all ORFs except the first one.
for seq_def in tuple(seqs.keys()):
seq_def_prefix = seq_def.split(' ')[0]
if seq_def_prefix.endswith('ORF001'):
max_title_a_len = max(len(seq_def_prefix),
max_title_a_len)
max_title_a_len += 2
max_run_id_len += 2
# --------------------------------------------------------------------
parallel_run_count = min(THREADS, len(ss_names))
def run_inter_pro_scan_parallel(ss):
if stat(aa_queries_files[ss]).st_size == 0:
return
run_inter_pro_scan(ss, assemblies, email, dir_prj_ips,
dir_cache_prj, parallel_run_count,
max_title_a_len, max_run_id_len)
# GFF3 files from kakapo and InterProScan 5 results JSON files
gff_from_json(ss, assemblies, dir_prj_ips,
dir_prj_transcripts_combined, prj_name)
Parallel(n_jobs=parallel_run_count, verbose=0,
require='sharedmem')(delayed(run_inter_pro_scan_parallel)(ss)
for ss in ss_names)
# Download CDS for NCBI protein queries ----------------------------------
print()
prot_cds_ncbi_files = OrderedDict()
def dnld_cds_for_ncbi_prot_acc_parallel(ss):
if stat(aa_queries_files[ss]).st_size == 0:
return
if ss not in prot_acc_user_filtered:
return
prot_cds_ncbi_files[ss] = opj(
dir_prj_transcripts_combined,
prj_name + '_ncbi_query_cds__' + ss + '.fasta')
if len(prot_acc_user_filtered[ss]) > 0:
dnld_cds_for_ncbi_prot_acc(ss, prot_acc_user_filtered[ss],
prot_cds_ncbi_files[ss], tax,
dir_cache_prj)
ss_names = tuple(sss.keys())
Parallel(n_jobs=2, verbose=0, require='sharedmem')(
delayed(dnld_cds_for_ncbi_prot_acc_parallel)(ss) for ss in ss_names)
# ------------------------------------------------------------------------
rmtree(dir_temp)
# ------------------------------------------------------------------------
rerun = input('\nRepeat ([y]/n)? ').lower().strip()
if rerun.startswith('y') or rerun == '':
print()
return False
else:
print('\nExiting...')
return True
def run_spades(se_fastq_files, pe_fastq_files, dir_spades_assemblies,
spades, dir_temp, ss, threads, ram):
if len(se_fastq_files) > 0 or len(pe_fastq_files) > 0:
if spades is None:
Log.err('SPAdes is not available. Cannot continue. Exiting.')
exit(0)
for se in se_fastq_files:
dir_results = opj(dir_spades_assemblies, se + '__' + ss)
fq_path = se_fastq_files[se]['vsearch_results_path' + '__' + ss]
se_fastq_files[se]['spades_assembly' + '__' + ss] = None
if ope(dir_results):
Log.msg('SPAdes assembly already exists:', se)
else:
make_dirs(dir_results)
Log.msg('Running SPAdes on:', se)
run_spades_se(spades,
out_dir=dir_results,
input_file=fq_path,
threads=threads,
memory=ram,
rna=True)
assmbl_path = opj(dir_results, 'transcripts.fasta')
if ope(assmbl_path):
count = len(read_fasta(assmbl_path, SEQ_TYPE_NT))
tr_str = ' transcripts.'
if count == 1:
tr_str = ' transcript.'
Log.msg('SPAdes produced ' + str(count) + tr_str, False)
se_fastq_files[se]['spades_assembly' + '__' + ss] = assmbl_path
else:
Log.wrn('SPAdes produced no transcripts.', False)
for pe in pe_fastq_files:
dir_results = opj(dir_spades_assemblies, pe + '__' + ss)
fq_paths = pe_fastq_files[pe]['vsearch_results_path' + '__' + ss]
pe_fastq_files[pe]['spades_assembly' + '__' + ss] = None
if ope(dir_results):
Log.msg('SPAdes assembly already exists:', pe)
else:
make_dirs(dir_results)
Log.msg('Running SPAdes on: ' + pe)
if osstat(fq_paths[0]).st_size > 0 and \
osstat(fq_paths[1]).st_size > 0:
run_spades_pe(spades,
out_dir=dir_results,
input_files=fq_paths,
threads=threads,
memory=ram,
rna=True)
else:
_ = opj(dir_temp, 'temp.fasta')
combine_text_files(fq_paths, _)
run_spades_se(spades,
out_dir=dir_results,
input_file=_,
threads=threads,
memory=ram,
rna=True)
osremove(_)
assmbl_path = opj(dir_results, 'transcripts.fasta')
if ope(assmbl_path):
count = len(read_fasta(assmbl_path, SEQ_TYPE_NT))
tr_str = ' transcripts.'
if count == 1:
tr_str = ' transcript.'
Log.msg('SPAdes produced ' + str(count) + tr_str, False)
pe_fastq_files[pe]['spades_assembly' + '__' + ss] = assmbl_path
else:
Log.wrn('SPAdes produced no transcripts.', False)
def run_inter_pro_scan(ss, assemblies, email, dir_prj_ips, dir_cache_prj,
parallel_run_count, max_title_a_len, max_run_id_len):
delay = 0.25
for a in assemblies:
if 'transcripts_aa_orf_fasta_file__' + ss not in a:
continue
aa_file = a['transcripts_aa_orf_fasta_file__' + ss]
if aa_file is None:
continue
assmbl_name = a['name']
json_dump_file_path = opj(dir_prj_ips,
assmbl_name + '_ann_ips__' + ss + '.json')
if ope(json_dump_file_path):
Log.inf('InterProScan results for assembly ' + assmbl_name + ', '
'search strategy ' + ss + ' have already been downloaded.')
continue
else:
Log.inf('Running InterProScan on translated ' + ss + ' from ' +
assmbl_name + '.')
seqs = seq_records_to_dict(read_fasta(aa_file, SEQ_TYPE_AA))
# Filter all ORFs except the first one.
for seq_def in tuple(seqs.keys()):
seq_def_prefix = seq_def.split(' ')[0]
if not seq_def_prefix.endswith('ORF001'):
del seqs[seq_def]
seqs = OrderedDict(
sorted(seqs.items(),
key=lambda x: x[0].split(' ')[1],
reverse=True))
run_id = ss + '_' + assmbl_name
_ = opj(dir_cache_prj, 'ips5_cache_done_' + run_id)
if ope(_):
with open(_, 'rb') as f:
jobs = pickle.load(f)
else:
jobs = job_runner(email=email,
dir_cache=dir_cache_prj,
seqs=seqs,
run_id=run_id,
parallel_run_count=parallel_run_count,
max_title_a_len=max_title_a_len,
max_run_id_len=max_run_id_len)
with open(_, 'wb') as f:
pickle.dump(jobs, f, protocol=PICKLE_PROTOCOL)
Log.inf('Downloading InterProScan results for ' + ss + ' in ' +
assmbl_name + '.')
all_ips_results = {}
# Nicer printing
for i, job in enumerate(jobs['finished']):
job_id = jobs['finished'][job]
titles_ab = split_seq_defn(job)
title_a = titles_ab[0]
progress = round(((i + 1) / len(jobs['finished'])) * 100)
progress_str = '{:3d}'.format(progress) + '%'
msg = (' ' * 12 + title_a.ljust(max_title_a_len) +
run_id.ljust(max_run_id_len) + progress_str.rjust(4) + ' ' +
job_id)
Log.msg(msg)
sleep(delay)
ips_json = result_json(job_id)
if ips_json is None:
continue
# ips_version = ips_json['interproscan-version']
ips_json = ips_json['results']
# These fields are set to 'EMBOSS_001' by default
# Delete them
del ips_json[0]['xref']
job_no_def = job.split(' ')[0]
all_ips_results[job_no_def] = ips_json
with open(json_dump_file_path, 'w') as f:
json.dump(all_ips_results, f, sort_keys=True, indent=4)
# Removes cached jobs file.
osremove(_)
def run_tblastn_on_reads(se_fastq_files, pe_fastq_files, aa_queries_file,
tblastn, blast_1_evalue, blast_1_max_hsps,
blast_1_qcov_hsp_perc, blast_1_best_hit_overhang,
blast_1_best_hit_score_edge, blast_1_max_target_seqs,
dir_blast_results_fa_trim, fpatt, ss, threads, seqtk,
vsearch, dir_cache_prj):
changed_blast_1 = False
if len(se_fastq_files) > 0 or len(pe_fastq_files) > 0:
print()
Log.inf('Running BLAST on reads:', ss)
if tblastn is None:
Log.err('tblastn is not available. Cannot continue. Exiting.')
exit(0)
if vsearch is None:
Log.err('vsearch is not available. Cannot continue. Exiting.')
exit(0)
if seqtk is None:
Log.err('seqtk is not available. Cannot continue. Exiting.')
exit(0)
cache_file = opj(dir_cache_prj, 'blast_1_settings_cache__' + ss)
pickled = dict()
settings = {
'blast_1_evalue': blast_1_evalue,
'blast_1_max_hsps': blast_1_max_hsps,
'blast_1_qcov_hsp_perc': blast_1_qcov_hsp_perc,
'blast_1_best_hit_overhang': blast_1_best_hit_overhang,
'blast_1_best_hit_score_edge': blast_1_best_hit_score_edge,
'blast_1_max_target_seqs': blast_1_max_target_seqs,
'queries': seq_records_to_dict(read_fasta(aa_queries_file,
SEQ_TYPE_AA))
}
Log.msg('evalue:', str(blast_1_evalue))
Log.msg('max_hsps:', str(blast_1_max_hsps))
Log.msg('qcov_hsp_perc:', str(blast_1_qcov_hsp_perc))
Log.msg('best_hit_overhang:', str(blast_1_best_hit_overhang))
Log.msg('best_hit_score_edge:', str(blast_1_best_hit_score_edge))
Log.msg('max_target_seqs:', str(blast_1_max_target_seqs))
print()
# FixMe: Expose in configuration files?
ident = 0.85
for se in se_fastq_files:
dir_results = opj(dir_blast_results_fa_trim, se)
blast_db_path = se_fastq_files[se]['blast_db_path']
fq_path = se_fastq_files[se]['filter_path_fq']
out_f = opj(dir_results, se + '__' + ss + '.txt')
out_f_fastq = out_f.replace('.txt', '.fastq')
out_f_fasta = out_f.replace('.txt', '.fasta')
se_fastq_files[se]['blast_results_path' + '__' + ss] = out_f_fasta
genetic_code = se_fastq_files[se]['gc_id']
if ope(out_f_fasta) and ope(cache_file):
with open(cache_file, 'rb') as f:
pickled = pickle.load(f)
if ope(out_f_fasta) and pickled == settings:
# Log.msg('The provided BLAST settings and query sequences did '
# 'not change since the previous run.')
Log.msg('BLAST results already exist:', se)
else:
changed_blast_1 = True
make_dirs(dir_results)
Log.msg('Running tblastn on: ' + basename(blast_db_path), ss)
run_blast(exec_file=tblastn,
task='tblastn',
threads=threads,
db_path=blast_db_path,
queries_file=aa_queries_file,
out_file=out_f,
evalue=blast_1_evalue,
max_hsps=blast_1_max_hsps,
qcov_hsp_perc=blast_1_qcov_hsp_perc,
best_hit_overhang=blast_1_best_hit_overhang,
best_hit_score_edge=blast_1_best_hit_score_edge,
max_target_seqs=blast_1_max_target_seqs,
db_genetic_code=genetic_code,
out_cols=BLST_RES_COLS_1)
Log.inf('Extracting unique BLAST hits using Seqtk:', ss)
keep_unique_lines_in_file(out_f)
seqtk_extract_reads(seqtk, fq_path, out_f_fastq, out_f)
seqtk_fq_to_fa(seqtk, out_f_fastq, out_f_fasta)
osremove(out_f)
osremove(out_f_fastq)
out_f_fasta_temp = out_f_fasta + '_temp'
copyfile(out_f_fasta, out_f_fasta_temp)
run_cluster_fast(vsearch, ident, out_f_fasta_temp, out_f_fasta)
osremove(out_f_fasta_temp)
for pe in pe_fastq_files:
dir_results = opj(dir_blast_results_fa_trim, pe)
blast_db_paths = pe_fastq_files[pe]['blast_db_path']
fq_paths = pe_fastq_files[pe]['filter_path_fq']
out_fs = [x.replace('@D@', dir_results) for x in fpatt]
out_fs = [x.replace('@N@', pe) for x in out_fs]
out_fs = [x.replace('@Q@', ss) for x in out_fs]
out_fs_fastq = [x.replace('.txt', '.fastq') for x in out_fs]
out_fs_fasta = [x.replace('.txt', '.fasta') for x in out_fs]
out_f_fasta = opj(dir_results, pe + '__' + ss + '.fasta')
pe_fastq_files[pe]['blast_results_path' + '__' + ss] = out_f_fasta
genetic_code = pe_fastq_files[pe]['gc_id']
if ope(out_f_fasta) and ope(cache_file):
with open(cache_file, 'rb') as f:
pickled = pickle.load(f)
if ope(out_f_fasta) and pickled == settings:
# Log.msg('The provided BLAST settings and query sequences did '
# 'not change since the previous run.')
Log.msg('BLAST results already exist:', pe)
else:
changed_blast_1 = True
make_dirs(dir_results)
pe_trim_files = zip(blast_db_paths, out_fs, fq_paths, out_fs_fastq,
out_fs_fasta)
for x in pe_trim_files:
Log.msg('Running tblastn on: ' + basename(x[0]), ss)
run_blast(exec_file=tblastn,
task='tblastn',
threads=threads,
db_path=x[0],
queries_file=aa_queries_file,
out_file=x[1],
evalue=blast_1_evalue,
max_hsps=blast_1_max_hsps,
qcov_hsp_perc=blast_1_qcov_hsp_perc,
best_hit_overhang=blast_1_best_hit_overhang,
best_hit_score_edge=blast_1_best_hit_score_edge,
max_target_seqs=blast_1_max_target_seqs,
db_genetic_code=genetic_code,
out_cols=BLST_RES_COLS_1)
Log.msg('Extracting unique BLAST hits using Seqtk:', ss)
keep_unique_lines_in_file(x[1])
seqtk_extract_reads(seqtk, x[2], x[3], x[1])
seqtk_fq_to_fa(seqtk, x[3], x[4])
osremove(x[1])
osremove(x[3])
combine_text_files(out_fs_fasta, out_f_fasta)
out_f_fasta_temp = out_f_fasta + '_temp'
copyfile(out_f_fasta, out_f_fasta_temp)
run_cluster_fast(vsearch, ident, out_f_fasta_temp, out_f_fasta)
osremove(out_f_fasta_temp)
for x in out_fs_fasta:
osremove(x)
with open(cache_file, 'wb') as f:
pickle.dump(settings, f, protocol=PICKLE_PROTOCOL)
return changed_blast_1