tblastn比对出来候选HSP区段，我们需要根据一定的基因长度范围来进行区域延伸去重叠，然后进行exonerate或者genewise建模。对HSP区域的延伸要考率基因的长度以及目标基因组scafflod or chromosome长度，不是一件容易的事情。
这里采用了dataclass以及改写slots存储数据方式，减少内存占用以及加快读取速度，attrgetter对列表字典结构排序，biopython里的SeqIO.parse感觉挺慢的，不过懒得重写了，用了据说更快的SimpleFastaParser来解析，实际测试下来速度确实更快，主要是用来存储scaffold或chromosome长度，以防延伸超出边界。
去重叠的原理在于先排序，然后判断前一区间的末尾是否小于后一区间开始，若为假则重叠，根据长度/得分来判断删除前一区间还是后一区间。
以下运行得到的结果仍然是blast的tabular格式(之后可以经过一些简单的shell命令处理，可转成bed格式，结合bedtools批量提取序列)。
注：如果你需要去重的格式不为blast tabular，简单的利用一些工具如awk/sed/perl/python/shell各种改变格式就好，只需要第二列的id，第九列的序列起始，第十列的序列结束，第十二列的得分有意义，作为排序用到的字段，其余字段都可缺省
from dataclasses import dataclass, replace
from operator import attrgetter
from Bio.SeqIO.FastaIO import SimpleFastaParser
import getopt
import sys
"""
用法:
#根据基因组文件，延伸上下游区域
example 1. region_tools.py -u <int> -d <int> -i <blast_tabular> -o extend.txt [-g <genome_file>]
#去除重叠区域，保留最长，并用-t指定筛去小于某长度的结果
example 2. region_tools.py -f -i extend.txt -t <region_length> -o filter.txt
#延伸+去重+根据得分保留
example 3. region_tools.py -u <int> -d <int> -f -s -i <blast_tabular> [-g <genome_file>] -o output.txt
"""
@dataclass
class Rec:
__slots__ = ("q_id", "s_id", "identity", "alignment_length", "mismatches",
"gap_openings", "q_start", "q_end", "s_start", "s_end",
"e_value", "bit_score")
q_id: str
s_id: str
identity: str
alignment_length: str
mismatches: str
gap_openings: str
q_start: str
q_end: str
s_start: int
s_end: int
e_value: str
bit_score: float
def filter_overlap(sort_data, score=False):
i = 0
if sort_data:
if sort_data[0].s_start <= sort_data[0].s_end:
start = "s_start"
end = "s_end"
else:
start = "s_end"
end = "s_start"
while i <= len(sort_data) - 2:
if getattr(sort_data[i+1], start) < getattr(sort_data[i], end) and \
sort_data[i+1].s_id == sort_data[i].s_id:
if not score:
if abs(sort_data[i+1].s_start - sort_data[i+1].s_end) > \
abs(sort_data[i].s_start - sort_data[i].s_end):
del sort_data[i]
else:
del sort_data[i+1]
else:
if sort_data[i+1].bit_score > sort_data[i].bit_score:
del sort_data[i]
else:
del sort_data[i+1]
else:
i += 1
class RegionIO:
def __init__(self, file_path):
self.file_path = file_path
self.sort_rf = []
self.sort_rv = []
self.recs_forward = []
self.recs_reverse = []
with open(self.file_path, "r") as f:
for rec in f:
rec = rec.strip().split("\t")
bit_score = float(rec[11])
rec_dict = Rec(
*rec[0:8],
int(rec[8]),
int(rec[9]),
rec[10],
int(bit_score) if bit_score == int(bit_score) else bit_score)
if rec_dict.s_start <= rec_dict.s_end:
self.recs_forward.append(rec_dict)
else:
self.recs_reverse.append(rec_dict)
def extend_region(self, up=0, down=0, genome_path=""):
if genome_path == "":
for n in range(len(self.recs_forward)):
start = self.recs_forward[n].s_start - up
end = self.recs_forward[n].s_end + down
self.recs_forward[n] = \
replace(self.recs_forward[n], s_start=start) \
if start >= 1 \
else replace(self.recs_forward[n], s_start=1)
self.recs_forward[n] = \
replace(self.recs_forward[n], s_end=end) \
if start >= 1 \
else replace(self.recs_forward[n],
s_end=abs(start)+end+1)
for n in range(len(self.recs_reverse)):
start = self.recs_reverse[n].s_end - up
end = self.recs_reverse[n].s_start + down
self.recs_reverse[n] = \
replace(self.recs_reverse[n], s_end=start) \
if start >= 1 \
else replace(self.recs_reverse[n], s_end=1)
self.recs_reverse[n] = \
replace(self.recs_reverse[n], s_start=end) \
if start >= 1 \
else replace(self.recs_reverse[n],
s_start=abs(start)+end+1)
else:
id_len = {}
with open(genome_path,"r") as in_handle:
for title, seq in SimpleFastaParser(in_handle):
id_len[title.split(" ")[0]] = len(seq)
for n in range(len(self.recs_forward)):
start = self.recs_forward[n].s_start - up
end = self.recs_forward[n].s_end + down
self.recs_forward[n] = \
replace(self.recs_forward[n], s_start=start) \
if start >= 1 \
else replace(self.recs_forward[n], s_start=1)
if end <= id_len[self.recs_forward[n].s_id]:
self.recs_forward[n] = \
replace(self.recs_forward[n], s_end=end) \
if start >= 1 \
else replace(self.recs_forward[n],
s_end=abs(start) + end + 1 if
abs(start) + end + 1 <=
id_len[self.recs_forward[n].s_id] else
id_len[self.recs_forward[n].s_id])
else:
self.recs_forward[n] = \
replace(self.recs_forward[n],
s_end=id_len[self.recs_forward[n].s_id])
self.recs_forward[n] = \
replace(self.recs_forward[n],
s_start=self.recs_forward[n].s_start
-(end-id_len[self.recs_forward[n].s_id])
if self.recs_forward[n].s_start
-(end-id_len[self.recs_forward[n].s_id]) >= 1 else 1)
for n in range(len(self.recs_reverse)):
start = self.recs_reverse[n].s_end - up
end = self.recs_reverse[n].s_start + down
self.recs_reverse[n] = \
replace(self.recs_reverse[n], s_end=start) \
if start >= 1 \
else replace(self.recs_reverse[n], s_end=1)
if end <= id_len[self.recs_reverse[n].s_id]:
self.recs_reverse[n] = \
replace(self.recs_reverse[n], s_start=end) \
if start >= 1 \
else replace(self.recs_reverse[n],
s_start=abs(start) + end + 1 if
abs(start) + end + 1 <=
id_len[self.recs_forward[n].s_id] else
id_len[self.recs_forward[n].s_id])
else:
self.recs_reverse[n] = \
replace(self.recs_reverse[n],
s_start=id_len[self.recs_reverse[n].s_id])
self.recs_reverse[n] = \
replace(self.recs_reverse[n],
s_end=self.recs_reverse[n].s_end
-(end - id_len[self.recs_reverse[n].s_id])
if (self.recs_reverse[n].s_end
-(end - id_len[self.recs_reverse[n].s_id])) >= 1 else 1)
def region_sort(self):
self.sort_rf = sorted(self.recs_forward,
key=attrgetter("s_id", "s_start", "s_end"))
self.sort_rv = sorted(self.recs_reverse,
key=attrgetter("s_id", "s_end", "s_start"))
def filter_threshold(self, threshold):
if int(threshold) > 0:
self.sort_rf = [record for record in self.sort_rf
if abs(record.s_start-record.s_end)+1 >= threshold]
self.sort_rv = [record for record in self.sort_rv
if abs(record.s_start-record.s_end)+1 >= threshold]
def write(self, output_path):
with open(output_path, "w") as o:
for rec in self.sort_rf+self.sort_rv:
o.write(f"{rec.q_id}\t{rec.s_id}"
f"\t{rec.identity}\t{rec.alignment_length}"
f"\t{rec.mismatches}\t{rec.gap_openings}"
f"\t{rec.q_start}\t{rec.q_end}"
f"\t{rec.s_start}\t{rec.s_end}"
f"\t{rec.e_value}\t{rec.bit_score}\n")
class StepError(Exception):
def __init__(self, error):
self.error = error
if __name__ == "__main__":
path, up, down, filt, genome, output, threshold, score = "", 0, 0, False, "", "", 0, False
try:
opts, args = getopt.getopt(sys.argv[1:], "-i:-g:-o:-u:-d:-f-t:-s")
for opt_name, opt_value in opts:
if opt_name == "-i":
path = opt_value
if opt_name == "-u":
up = int(opt_value)
if opt_name == "-d":
down = int(opt_value)
if opt_name == "-f":
filt = True
if opt_name == "-t":
threshold = int(opt_value)
if opt_name == "-g":
genome = opt_value
if opt_name == "-o":
output = opt_value
if opt_name == "-s":
score = True
except getopt.GetoptError as e:
for error in e.args:
print("".join(error))
results = RegionIO(path)
if up or down:
if genome:
results.extend_region(up=up, down=down, genome_path=genome)
else:
results.extend_region(up=up, down=down)
results.region_sort()
if filt:
filter_overlap(results.sort_rf, score)
filter_overlap(results.sort_rv, score)
if threshold > 0:
results.filter_threshold(threshold)
if output:
results.write(output)
else:
results.write(path+"_region_tools")
