In-silico downsizing to estimate the DNA input

Question

[APJ]

Hi,

Given a fastq file from 50ng data, I could find all the reference variants from the variant calling results. Is it possible to test in silico downsizing of fastq data, to see what the minimal DNA amount would be to not lose our reference variants?

Any thoughts on this?

Thank you!

[ATpoint]

I guess all you can do is check how coverage differences change variant calls, but I doubt that you can meaningfully simulate different DNA amounts as this is dependent on the kit and the number of PCR cycles, so you would need data for different starting amounts and then make a model based on these data.

Answer 1

[5heikki]

Why not?

You can do e.g. this:

paste -d $'\t' - - - - <file.fq | shuf -n "$NUMBER" | awk 'BEGIN{FS="\t";OFS="\n"}{print $1,$2,$3,$4}' > out.fq

Where "$NUMBER" is the number of reads you want in your output. If you want the shuf to be deterministic or e.g. have the chance to including the same read more than once then see man shuf