jellyfish count discrepancy between results obtained with and without -C option

Question

[ashenflower]

I ran jellyfish on my Illumina reads, first using the -C options, and then without, on the same data. Those are the commands I used:

With -C option:

time jellyfish count -m 21 -s 100M -t 15 -C <(zcat file_1.fastq.gz file_2.fastq.gz)

Without -C option:

time jellyfish count -m 21 -s 100M -t 15 <(zcat file_1.fastq.gz file_2.fastq.gz)

In the first case I obtained 1.684.382.436 distinct k-mers, while in the second case I obtained 2.205.041.740, so only 520.659.304.

How is it possible? I was expecting the number of distinct k-mers in the second case (no -C) to be around twice the number in the first case (coverage is, approximately, x53)

Answer 1

[Rob]

This is not at all unexpected. The number without -C will count the kmers as they appear in the input. To have this number be twice what you get with -C, then every kmer would have to appear in both its forward and reverse complement orientation. With 50X coverage, you might expect this with most (not all) kmers in the true underlying genome. However, many erroneous kmers (from sequencing error), are likely to occur only once, so you won't see them in both orientations, even at high coverage.

[ashenflower]

Thank you very much!

[onestop_data]

Well said!