Mastering Python for Bioinformatics

Chapter 18. FASTX Sampler: Randomly Subsampling Sequence Files

Sequence datasets in genomics and metagenomics can get dauntingly large, requiring copious time and compute resources to analyze. Many sequencers can produce tens of millions of reads per sample, and many experiments involve tens to hundreds of samples, each with multiple technical replicates resulting in gigabytes to terabytes of data. Reducing the size of the input files by randomly subsampling sequences allows you to explore data more quickly. In this chapter, I will show how to use Python’s random module to select some portion of the reads from FASTA/FASTQ sequence files.

You will learn about:

• Nondeterministic sampling

Getting Started

The code and tests for this exercise are in the 18_fastx_sampler directory. Start by copying the solution for a program called sampler.py:

$ cd 18_fastx_sampler/
$ cp solution.py sampler.py

The FASTA input files for testing this program will be generated by the synth.py program you wrote in Chapter 17. If you didn’t finish writing that program, be sure to copy the solution to that filename before executing make fasta to create three FASTA files with 1K, 10K, and 100K reads, each between 75 and 200 bp in length, with filenames of n1k.fa, n10k.fa, and n100k.fa, respectively. Use seqmagique.py to verify that the files are correct:

$ ../15_seqmagique/seqmagique.py tests/inputs/n1* name min_len max_len avg_len num_seqs tests/inputs/n100k.fa 75 200 136.08 100000 tests/inputs/n10k.fa ...