Jellyfish & Paired end reads

Question

[joshua.j.wright]

I am currently going through kmer analysis of some 2x150 PE sequencing that I carried out for a first year PhD project for whole genome sequencing.

I carried out five runs so have 10 paired end files from one individual. As this is the case, do all reads need merging/concatenating before going through jellyfish or should only one of the paired end reads be merged and put through that way?

Alternatively - as jellyfish reads all canonical Kmers would just putting the forward read of one of the runs be enough as it would read it forward and reverse anyway?

Thanks in advance!

[Darked89]

You may need to do some QC first before doing kmer counting. I would look for Illumina adapters/multiplexing sequences just in case. Also you want to know if you have some inserts shorter than 300bp so the ends of r1 and r2 may overlap.

[joshua.j.wright]

QC has already been carried out! There is no adapters in there and reads have been quality trimmed

[GenoMax]

I carried out five runs so have 10 paired end files from one individual.

Are these technical replicates of sequencing a single library or are these five independent libraries?

[joshua.j.wright]

So there is two libraries, three replicates of one, two replicates of the other.

Answer 1

[Darked89]

Before starting jellyfish you may take a look at the benchmarks 2018 benchmarks and try to estimate if k-mer counting for your organism and data set combo could be processed by jellyfish in a reasonable time.

The other point from the benchmark article above is that IO can be a bottleneck. For that reason I would cluster/reorder the reads (individual paired FASTQs) using clumpify from BBMap and run some benchmarks maybe starting with one or few r1 FASTQ files.
I would hope that ordering reads with clumpify should improve lookup times for a given k-mer but this is likely algorithm dependent.