BBduk for NexteraXT

[arivers]

I'm using BBduk 38.00 for adaptor trimming on NexteraXT libraries multiplexed with i5 and i7 indices and sequenced on an NextSeq500 2x150:

bbduk.sh in=sample.fq.gz ref=adapters.fa bhist=bhist.txt k=23 hdist=1 ktrim='r' overwrite=t mink=8 tbe tbo ftm=5

When I plot the bhist output from BBduk (A, C, G, T, N % at each position on read one then read2), I get strong GC and AT precent divergence over the first 16-18 bp of read 1 and read 2.

It looks like I'm missing some forward primer or transposase sequence and, in fact, the most common bases are similar to the read 2 Nextera transposase sequence.

Here is an example plot:
https://photos.app.goo.gl/hH2UjJ9iErPbC5nP9

[GenoMax]

Have you come across this blog post?

What you are observing is the "bias" observed similar to random primed libraries that is present in nextera libraries.

[arivers]

I knew that was an issue for random hexamer primed RNAseq libraries and I've heard that tagmentation sites were non-random. I usually work with Truseq/kapa data so this is my first time actually seeing how non-random things are.

Using BBduk to left trim with a hamming distance of 2 or 3 reduces, or at least masks the positional sequence bias.

Glad to know the consensus is that it's not a problem.

[GenoMax]

You should align the data without doing any additional manipulations beyond removal of adapters. Data should align normally.