I want to hear some bioinformatics war stories! What was the most weird bug or sequencing artifact you had to debug? I'll start: Did you know that changing the number of threads used by bwa-mem can influence your BAM output? - ThreadSky

About ThreadSky

lucasvandijk.nl • 104 days ago

I want to hear some bioinformatics war stories! What was the most weird bug or sequencing artifact you had to debug?

I'll start: Did you know that changing the number of threads used by bwa-mem can influence your BAM output?

Comments

lucasvandijk.nl•104 days ago

Consider the case when aligning reads from a whole metagenomic sequencing sample to a specific bacterial reference. For example, let's align the reads to an E. coli reference to characterize E. coli strains in that sample.

lucasvandijk.nl•104 days ago

E. coli in the human gut is typically at low abundance (<1%). This means that for a sample sequenced to a depth of 2Gb, you'll get, at best, around 4x coverage of a 5 Mb E. coli genome. Importantly, sample reads comprise a mixture of E. coli reads and reads from other species in the gut microbiome.

lucasvandijk.nl•104 days ago

When bwa-mem aligns reads, it searches for the best location in the reference to align R1 and its paired-end mate R2. It cleverly takes into account the DNA fragment length distribution to ensure R1 and R2 are not several kilobases apart.

lucasvandijk.nl•104 days ago

To prevent reading the FASTQ files twice—once to estimate the fragment length distribution and once to actually align reads—bwa mem loads reads from the FASTQs in batches (by default, around 60,000 per batch).

lucasvandijk.nl•104 days ago

BWA aligns the batch reads to the reference and uses the unambiguous, in correct FR orientation, alignments to estimate the fragment size mean and std for this batch. It uses these statistics to further rescue/disambiguate missing or ambiguous alignments.

lucasvandijk.nl•104 days ago

Here's the kicker: if the species of interest in a metagenomic sample is at low abundance, occasionally so few reads will successfully align to the reference that BWA cannot estimate the fragment size mean and std.

Posting Rules

Be respectful to others
No spam or self-promotion
Stay on topic
Follow Bluesky's terms of service

Comments

Posting Rules

Reply