print("# negative .fa sequences: %i" % (c_neg_fa))
# Check additional files.
if args.opt_pos_fa:
assert args.opt_neg_fa, "--opt-pos but no --opt-neg given"
if args.opt_neg_fa:
assert args.opt_pos_fa, "--opt-neg but no --opt-pos given"
# Check for lowercase only sequences, which cause GP to crash.
error_mess = (
"input sequences encountered containing "
"only lowercase characters or lowercase characters in between "
"uppercase characters. Please provide either all uppercase "
"sequences or sequences containing uppercase regions surrounded "
"by lowercase context regions for structure calculation (see "
"viewpoint concept in original GraphProt publication "
"for more details)")
seqs_dic = gplib.read_fasta_into_dic(args.in_pos_fa)
bad_ids = gplib.check_seqs_dic_format(seqs_dic)
assert not bad_ids, "%s" % (error_mess)
seqs_dic = gplib.read_fasta_into_dic(args.in_neg_fa)
bad_ids = gplib.check_seqs_dic_format(seqs_dic)
assert not bad_ids, "%s" % (error_mess)
if args.opt_pos_fa:
seqs_dic = gplib.read_fasta_into_dic(args.opt_pos_fa)
bad_ids = gplib.check_seqs_dic_format(seqs_dic)
assert not bad_ids, "%s" % (error_mess)
if args.opt_neg_fa:
seqs_dic = gplib.read_fasta_into_dic(args.opt_neg_fa)
bad_ids = gplib.check_seqs_dic_format(seqs_dic)
assert not bad_ids, "%s" % (error_mess)
# If parop .fa files given.
assert "linux" in sys.platform, "please use Linux"
# Check tool availability.
assert gplib.is_tool("GraphProt.pl"), "GraphProt.pl not in PATH"
# Check file inputs.
assert os.path.exists(
args.in_fa), 'input .fa file "%s" not found' % (args.in_fa)
assert os.path.exists(
args.in_model), 'input .model file "%s" not found' % (args.in_model)
assert os.path.exists(
args.in_params), 'input .params file "%s" not found' % (args.in_params)
# Count .fa entries.
c_in_fa = gplib.count_fasta_headers(args.in_fa)
assert c_in_fa, 'input .fa file "%s" no headers found' % (args.in_fa)
print("# input .fa sequences: %i" % (c_in_fa))
# Read in FASTA sequences to check for uppercase sequences.
seqs_dic = gplib.read_fasta_into_dic(args.in_fa)
# Check for lowercase only sequences, which cause GP to crash.
error_mess = (
"input sequences encountered containing "
"only lowercase characters or lowercase characters in between "
"uppercase characters. Please provide either all uppercase "
"sequences or sequences containing uppercase regions surrounded "
"by lowercase context regions for structure calculation (see "
"viewpoint concept in original GraphProt publication "
"for more details)")
if args.ws_pred:
bad_ids = gplib.check_seqs_dic_format(seqs_dic)
assert not bad_ids, "%s" % (error_mess)
c_uc_nt = gplib.seqs_dic_count_uc_nts(seqs_dic)
assert c_uc_nt, ("no uppercase nucleotides in input .fa sequences. "