Question

I have a raw, unaligned fastq.gz file that I am trying to preprocess using Biopython before alignment. I would ultimately like to remove low quality reads, trim polyA tails, trim adapters using fuzzy matching, and finally remove reads that do not satisfy a length requirement after all said preprocessing. It would also be neat to specify how many reads satisfy the filtering criteria at each step. I have been playing around with this Biopython script but have had little success.

# coding: utf-8

'''
process_reads.py
March 31, 2016

Convert FQ files to unaligned and tagged BAMs
I: compressed FQ
O: compressed, filtered FQ
'''

import gzip, statistics
from Bio import SeqIO, pairwise2

def get_stats(reads):
    sizes = [len(r) for r in reads]
    print("Total reads: %i" % len(sizes))
    print("Mean read length: %i" % statistics.mean(sizes))
    print("Max. read length: %i" % max(sizes))
    print("Min. read length: %i" % min(sizes))

def quality_filter(reads, qual):
    return (r for r in reads if min(r.letter_annotations["phred_quality"]) >= qual)

def trim_polyA(records, numA, minLen):
    for record in records:
        if len(record) < minLen: continue
        record = record.seq.split("A"*numA, 1)[0]
        yield record

def _remove_adaptor(seq, region, right_side=True):
    if right_side:
        try:
            pos = seq.find(region)
        # handle Biopython SeqRecords
        except AttributeError:
            pos = seq.seq.find(region)
        return seq[:pos]
    else:
        try:
            pos = seq.rfind(region)
        # handle Biopython SeqRecords
        except AttributeError:
            pos = seq.seq.rfind(region)
        return seq[pos+len(region):]

def trim_adaptor(seq, adaptor, num_errors, right_side=True):
    gap_char = '-'
    exact_pos = str(seq).find(adaptor)
    if exact_pos >= 0:
        seq_region = str(seq[exact_pos:exact_pos+len(adaptor)])
        adapt_region = adaptor
    else:
        seq_a, adaptor_a, score, start, end = pairwise2.align.localms(str(seq),
                                                                      str(adaptor),
                                                                      5.0, -4.0, -9.0, -0.5,
                                                                      one_alignment_only=True,
                                                                      gap_char=gap_char)[0]
        adapt_region = adaptor_a[start:end]
        seq_region = seq_a[start:end]
    matches = sum((1 if s == adapt_region[i] else 0) for i, s in enumerate(seq_region))
    # too many errors -- no trimming
    if (len(adaptor) - matches) > num_errors:
        return seq
    # remove the adaptor sequence and return the result
    else:
        return _remove_adaptor(seq, seq_region.replace(gap_char, ""),
                right_side)

def process_reads(fq, qual, adapt, numA, minLen):
    with gzip.open(fq) as f:
        rawReads = SeqIO.parse(f, "fastq")
        # get_stats(rawReads) # When I run this, everything downstream fails..
        qualFil = quality_filter(rawReads, qual) # I think this work fine.
        trimmedPoly = trim_polyA(qualFil, numA, minLen)
        trimmedAdap = trim_adaptor(trimmedPoly, adapt, 2)

        # count = SeqIO.write(trimmedAdap, "good_quality.fastq", "fastq")
        # print(count)

# TEST PROCESSING
fq = "test/TAAGGCGA_2.fq.gz"
process_reads(fq, qual=50, adapt="AAGCAGTGGTATCAACGCAGAGTGAATGGG", numA=6, minLen=20)

I believe the quality filter and polyA trimming works correctly but I cannot seem to get the adapters to cut. I have also written a function called get_stats that is supposed to return the average length and total reads. I would appreciate any help!

Answer 1

我想知道right_side始终设置为True。因此，您的_remove_adaptor方法只会尝试从右侧删除适配器：

...
if right_side:
        try:
            pos = seq.find(region)
        # handle Biopython SeqRecords
        except AttributeError:
            pos = seq.seq.find(region)
        return seq[:pos]

但是如果你从右边切掉polyA（保留序列的左边部分），你还会在右边有一个适配器序列吗？我的猜测是你要从左侧切割适配器。

一些示例读取会有所帮助。

BioPython: Processing raw RNAseq reads (quality filtering and trimming)

1 个答案: