marcelm

#!/usr/bin/env python3
"""Quality trimming using a running sum from the 5' to 3' end"""

import sys
from argparse import ArgumentParser
import dnaio


def qual_trim_index(qualities_ascii, threshold):
    qualities = [ord(c) - 33 for c in qualities_ascii]

marcelm

# Split reads in a FASTQ file at adapter occurrences
#
# Run:
#  cutadapt -O 100 --times=1000 -g MYADAPTERSEQ --info-file=info.txt -o /dev/null reads.fastq.gz
#
# Then:
#  awk -F "\t" -f split.awk info.txt | gzip > split.fastq.gz


# Relevant info file fields:

marcelm

#!/usr/bin/env python3
"""
Run with:
    fasta2fastq < in.fasta > out.fastq
"""
import dnaio
import sys
with dnaio.open(sys.stdin.buffer) as inf:
    with dnaio.open(sys.stdout.buffer, mode="w", fileformat="fastq") as outf:
        for record in inf:

marcelm

#!/bin/bash

# This script creates both
# - environment.osx.lock.yml and
# - environment.linux.lock.yml
# regardless of the operating system it is running on. The trick is
# temporarily setting the subdir and subdirs keys in .condarc to
# what would be appropriate for the other operating system.
#
# It assumes that there exists a (manually managed) environment.yml file

marcelm

#!/bin/bash

# A workaround for an issue with Nextflow (which may actually be a bash bug),
# see <https://github.com/SciLifeLab/Sarek/issues/420>
#
# The problem is that Nextflow does not notice that a job has finished and
# hangs indefinitely.
#
# This script looks for zombie processes that are children of a script named
# .command.stub, and kills that script. This seems to let the pipeline continue

marcelm

#!/usr/bin/env python3
"""
Mask low-quality bases in a FASTQ file with 'N'.
Adjust cutoff_front and cutoff_back below to use
different thresholds (currently: 20 at 5' end,
0 at 3' end).

Usage:
  python3 qualmask.py input.fastq.gz > output.fastq

marcelm

#!/bin/bash
set -euo pipefail
if [ $# -ne 1 -o x$1 == x-h -o x$1 == x--help ]; then
    echo \
"Usage:

    samtools sort -O bam -T prefix ... | bambai BAMPATH

Read a sorted BAM file from standard input, write it to BAMPATH and
index it at the same time (creating BAMPATH.bai)."

marcelm

from pysam import AlignmentFile
from pyfaidx import Fasta

def has_mismatch_in_interval(reference, bamfile, chrom, start, end):
    """
    Return whether there is a mismatch in the interval (start, end) in any read mapping to the given chromosome.

    reference -- a pyfaidx.Fasta object or something that behaves similarly
    """
    for column in bamfile.pileup(chrom, start, end):

marcelm

"""
Plot multiple figures into a single PDF with matplotlib, using the
object-oriented interface.
"""
from matplotlib.backends.backend_pdf import FigureCanvasPdf, PdfPages
from matplotlib.figure import Figure
import numpy as np

with PdfPages('multi.pdf') as pages:
	for i in range(10):

marcelm

#!/usr/bin/env python3
"""
Running this script is (intended to be) equivalent to running the following Snakefile:

include: "pipeline.conf"  # Should be an empty file

shell.prefix("set -euo pipefail;")

rule all:
    input: