How do you write a .gz fastq file with Biopython?

Question

How do you write a .gz (or .bgz) fastq file using Biopython?

I'd rather avoid a separate system call.

The typical way to write an ASCII .fastq is done as follows:

for record in SeqIO.parse(fasta, "fasta"):
    SeqIO.write(record, fastq, "fastq")

The record is a SeqRecord object, fastq is the file handle, and "fastq" is the requested file format. The file format may be fastq, fasta, etc., but I do not see an option for .gz.

Here is the SeqIO API.

Answer 1

I'm not sure I'm doing it the best way, but here is an example where I read a compressed gzip fastq file and write the records in block gzip fastq:

from Bio import SeqIO, bgzf
# Used to convert the fastq stream into a file handle
from io import StringIO
from gzip import open as gzopen

records = SeqIO.parse(
    # There is actually simpler (thanks @peterjc)
    # StringIO(gzopen("random_10.fastq.gz").read().decode("utf-8")),
    gzopen("random_10.fastq.gz", "rt"),
    format="fastq")

with bgzf.BgzfWriter("test.fastq.bgz", "wb") as outgz:
    SeqIO.write(sequences=records, handle=outgz, format="fastq")

Answer 2

Here is how I was able to export to a standard gzipped FASTQ file using Biopython. Basically, instead of using SeqIO.write, I directly called the .format method of the SeqRecord object. The example code below imports a gzipped FASTQ file, removes reads that do not contain a G in positions 7, 8, and 9, and writes the results to a gzipped FASTQ file.

from Bio import SeqIO
import gzip

path_in = "path/to/in.fastq.gz"
path_out = "path/to/out.fastq.gz"
handle_in = gzip.open(path_in, "rt")
handle_out = gzip.open(path_out, "wt")

fq = SeqIO.parse(handle_in, "fastq")
for read in fq:
    # Only export reads that have a G in positions 7, 8,
    # and 9
    if read.seq[6:9] == "GGG":
        handle_out.write(read.format("fastq"))

handle_in.close()
handle_out.close()

Using the code example from the original question that converts a FASTA file to a FASTQ file, it would look something like the following:

fastq = gzip.open("path/to/out.fastq.gz", "wt")
for record in SeqIO.parse(fasta, "fasta"):
    fastq.write(record.format("fastq"))
fastq.close()

Caveat: This only works for Python 3. Python 3 differs from Python 2 in how it imports data from binary files (source). In fact, ideally you should be using Python 3.4 or greater since Biopython has compatibility issues with Python 3.3 (source).

Answer 3

As far as I know you can not write compressed fastq on the fly. You will have to write the entire fastq and then compress it.