def directed_local_alignment(proteome_file, ref_seq_file, proteome_name,
workdir, mode, source, strand,
per_proteome_sequences):
"""
This function does the main lifting, running Supermatcher or BLAST on your proteomes/genomes.
"""
#open result file
alignment_file = os.path.join(workdir, "curr_alignment.aln")
#If Supermatcher was chosen as the alignment algorithm
if mode.lower() == "supermatcher":
#Unlike BLAST, Supermatcher only searches one strand, so we create a temp file to use as the supermatcher
#bsequence, and we can write into it either the positive or negative strand depending on the "strand" parameter
records = open_proteome(proteome_file)
if not records: return []
if not strand:
records = [
i.reverse_complement(id=True,
name=True,
description=True,
features=True,
annotations=True,
letter_annotations=True,
dbxrefs=True) for i in records
]
#write file
proteome_fasta_file = os.path.join(workdir, "curr_proteome.fasta")
SeqIO.write(records, proteome_fasta_file, "fasta")
#Build the matching command
if source.lower() == "protein" or source.lower() == "proteome":
matrix = "EBLOSUM62" #AA matrix
elif source.lower() == "nucleotide" or source.lower() == "genome":
matrix = "EDNAFULL"
#run supermatcher
cmd = SuperMatcherCommandline(asequence=ref_seq_file,
bsequence=proteome_fasta_file,
gapopen=10,
gapextend=0.5,
datafile=matrix,
outfile=alignment_file)
#Excecute the command
stdout, stderr = cmd()
#Parse the resulting alignments
alignments = []
try:
#Create list of MultipleSeqAlignment objects representing the Supermatcher results (list may be empty)
align_seq_list = list(AlignIO.parse(
alignment_file,
"amir_emboss")) #List of MultipleSeqAlignment objects
#Iterate through list, only the first alignment will be used if per_proteome_sequences==1 and all alignments
#will be used if per_proteome_sequences==None
for number, alignment in enumerate(
align_seq_list[0:per_proteome_sequences]):
#get the alignent
align_seq = alignment[
1] #SeqRecord objects, [0] is query and [1] is sbjct
#remove gaps
align_seq._set_seq(align_seq.seq.ungap("-"))
#get name
if (not per_proteome_sequences or per_proteome_sequences > 1
) and len(align_seq_list) > 1:
usable_name = proteome_name + "_" + str(number)
else:
usable_name = proteome_name
align_seq.name = usable_name
align_seq.id = usable_name
#finalize parsing
score, identity_percentage = parse_supermatcher_result(
alignment_file, number)
#add to list
alignments.append(
(score, align_seq, identity_percentage, 0,
1)) #1 is given arbitrarily as gene_percentage
#alignments is a list of tuples with each element being (score, align_seq, identity_percentage)
#detlete temp files
os.remove(alignment_file)
os.remove(proteome_fasta_file)
#return
return alignments
except ValueError or IndexError:
raise NoMatchForSeqException(proteome_fasta_file, ref_seq_file)
#If BLAST was chosen as the alignment algorithm
elif mode.upper() == "BLAST":
#if file is called XXXX.file_type.gz, db name should be XXXX
if proteome_file.endswith(".gz"):
db_name_temp = ".".join(proteome_file.split(".")[:-2])
#if file is called XXXX.file_type, db name should be XXXX
else:
db_name_temp = ".".join(proteome_file.split(".")[:-1])
#define dir
directory = os.path.dirname(proteome_file)
directory_files = os.listdir(directory)
#iterate through files and find database file
for dir_file in directory_files:
#determine type of die
if os.path.basename(
db_name_temp) in dir_file and ".nhr" in dir_file:
db_name = directory + "/" + dir_file.split(".nhr")[0]
elif os.path.basename(
db_name_temp) in dir_file and ".phr" in dir_file:
db_name = directory + "/" + dir_file.split(".phr")[0]
#open file
records = open_proteome(proteome_file)
if not records: return []
#Build matching command
if source.lower() == "protein" or source.lower() == "proteome":
cmd = NcbiblastpCommandline(query=ref_seq_file,
db=db_name,
out=alignment_file,
outfmt=5)
elif source.lower() == "nucleotide" or source.lower() == "genome":
cmd = NcbiblastnCommandline(query=ref_seq_file,
db=db_name,
out=alignment_file,
outfmt=5,
task="blastn")
#cmd = NcbiblastnCommandline(query=ref_seq_file, db=db_name, out=alignment_file, outfmt=5)
#Execute command
stdout, stderr = cmd()
#Open result
try:
result_handle = open(alignment_file)
blast_record = list(
NCBIXML.parse(result_handle))[0] #BLAST record object
result_handle.close()
except ValueError:
raise NoMatchForSeqException(proteome_file, ref_seq_file)
#Parse resulting alignments
alignments = []
try:
#Iterate through list, only the first alignment will be used if per_proteome_sequences==1 and all alignments
#will be used if per_proteome_sequences==None
for number, alignment in enumerate(
blast_record.alignments[0:per_proteome_sequences]):
hsp = alignment.hsps[
0] #HSP contains all the details about the alignment
sequence = hsp.sbjct
score = hsp.score
evalue = hsp.expect
identities = hsp.identities
query_length = blast_record.query_letters
align_length = hsp.align_length
#calculate percentages
identity_percentage = float(identities) / align_length
if (not per_proteome_sequences or per_proteome_sequences > 1
) and len(blast_record.alignments) > 1:
name = proteome_name + "_" + str(number)
else:
name = proteome_name
#length percentage
percentage = float(align_length) / query_length
#convert to SeqRecord0
align_seq = SeqRecord(Seq(sequence, IUPAC.protein),
id=name,
name=name,
description=name)
align_seq._set_seq(align_seq.seq.ungap("-")) #Remove the gaps
#score, identity_percentage = parse_blast_result(alignment_file)
alignments.append((score, align_seq, identity_percentage,
evalue, percentage))
#alignments is a list of tuples with each element being (score, align_seq, identity_percentage, evalue, length_percentage)
except IndexError: #If alignments are empty, doesn't actually do anything
sequence = ""
score = 0
identity_percentage = 0
#remove temp file
os.remove(alignment_file)
#return
return alignments
#not BLAST or Supermatcher!
else:
raise BaseException(
"Only Supermatcher and BLAST modes are currently supported.")
