Edit this page on GitHub

Split large file

Problem

With modern sequencing technologies it has become relatively cheap and easy to generate very large datasets. In fact, there are times when one can have too much data in one file, online resources like PLAN or TMHMM limit the size of users queries. In such cases it useful to be able to split a sequence file into a set of smaller files, each containing a subset of original file’s sequences.

Solution

There are many possible ways to solve this general problem, this recipe uses a generator function to avoid having all the data in memory at once.

def batch_iterator(iterator, batch_size):
    """Returns lists of length batch_size.

    This can be used on any iterator, for example to batch up
    SeqRecord objects from Bio.SeqIO.parse(...), or to batch
    Alignment objects from Bio.AlignIO.parse(...), or simply
    lines from a file handle.

    This is a generator function, and it returns lists of the
    entries from the supplied iterator.  Each list will have
    batch_size entries, although the final list may be shorter.
    """
    entry = True  # Make sure we loop once
    while entry:
        batch = []
        while len(batch) < batch_size:
            try:
                entry = iterator.next()
            except StopIteration:
                entry = None
            if entry is None:
                # End of file
                break
            batch.append(entry)
        if batch:
            yield batch

Here is an example using this function to divide up a large FASTQ file, SRR014849_1.fastq (from this compressed file):

from Bio import SeqIO

record_iter = SeqIO.parse(open("SRR014849_1.fastq"),"fastq")
for i, batch in enumerate(batch_iterator(record_iter, 10000)):
    filename = "group_%i.fastq" % (i + 1)
    with open(filename, "w") as handle:
        count = SeqIO.write(batch, handle, "fastq")
    print("Wrote %i records to %s" % (count, filename))

And the output:

Wrote 10000 records to group_1.fastq
Wrote 10000 records to group_2.fastq
Wrote 10000 records to group_3.fastq
Wrote 10000 records to group_4.fastq
Wrote 10000 records to group_5.fastq
Wrote 10000 records to group_6.fastq
...
Wrote 7251 records to group_27.fastq

You can modify this recipe to use any input and output formats supported by Bio.SeqIO, for example to break up a large FASTA file into units of 1000 sequences:

from Bio import SeqIO

record_iter = SeqIO.parse(open("large.fasta"),"fasta")
for i, batch in enumerate(batch_iterator(record_iter, 1000)):
    filename = "group_%i.fasta" % (i + 1)
    with open(filename, "w") as handle:
        count = SeqIO.write(batch, handle, "fasta")
    print("Wrote %i records to %s" % (count, filename))

How it works

It is possible to use list (SeqIO.parse(...)) to read the entire contents of a file into memory then write slices of the list out as smaller files. For large files (like the ones this recipe is about) that would take up a big hunk of memory, instead we can define a generator function, batch_iterator(), that loads one record at a time then appends it to a list, repeating the process until the list containins one file’s worth of sequences.

With that function defined it’s a matter of giving it an iterator (in this case a SeqIO.parse(...) instance and writing out the batching of records it produces.

Note that you can get close to this functionality with the built in Python function itertools.islice(record_iter, batch_size), but this gives an empty file at the end if the total count is an exact multiple of the batch size.