Illumina fastq lane output

Question

[sacha]

When I convert bcl to fastq (with bcl2fastq), by default I get several fastq file by sample AND by lane. For instance :

PAN150427_S17_L001_R1_001.fastq
PAN150427_S17_L002_R1_001.fastq
PAN150427_S17_L003_R1_001.fastq
PAN150427_S17_L004_R1_001.fastq

Is there any reason to not merge all lanes into one by default ? From this workflow http://www.htslib.org/workflow/, it seems they merge fastq lane after alignment. Is it really necessary ?

Answer 1

[GenoMax]

Keeping them separate allows you to work on the files in parallel speeding things up. You could merge the results files at a later step.

You could use --no-lane-splitting option with bcl2fastq v.2.18.x to get a single file per sample, if you prefer that.

Answer 2

[igor]

Some reasons for not merging:

• You (and anyone who might later work with your data) always knows how many lanes you ran. Although this may sound ridiculous for those who are keeping good records, it's a recurrent problem for many labs.
• If one of the lanes failed or had some issues, it's easy to take those sequences out.