Collapsing my paired-end reads before or after mapping ?

Question

[yolek64754]

Hi everyone,

I am about to map some Paired-End (PE) reads (aDNA) and I don't know what is the pro and cons about those two methods:

1. Run fastp (to remove the left-over adapters) -> map my PE dataset (using BWA) -> single BAM -> remove duplicates using Picard -> Mapping Quality filtering
2. Run Clip&Merge from EAGER (that collapse reads and trim adapters, specialised for aDNA) -> map my new SE dataset (using BWA) -> remove duplicates using Picard -> Mapping Quality filtering

My question is; is there any advantage for any of those two methods ?

Thanks a lot.

[GenoMax]

aDNA = amplicon? If EAGER is a special pipeline for such data then use that.

[yolek64754]

My bad, should not use abbreviations. aDNA stand for ancient DNA, 30-120bp damaged reads.

[GenoMax]

I would still recommend using EAGER since it seems to be designed for the special type of data you have.

[Friederike]

seconding Genomax on this -- why would you not want to use EAGER? Are there any specific reasons why you think alternative ways might be better suited for the task at hand?

[yolek64754]

I meant, correct me if I am wrong but if we collapse reads before mapping, we can loose some information from sequencing errors ? I aDNA most of the time the reads are short so the reads are fully sequenced in Left and Right.

[GenoMax]

EAGER must be taking that into account if you are following a standard protocol. That said every dataset has its own characteristics. You may need to start with #2 and if that does not work then try #1. I would still suggest this route.

[Friederike]

This would be the point where you could dig into the documentation of EAGER to find out how the collapsing is done. Presumably they are leveraging the fact that both reads tend to cover the same piece of a fragment.

[Gabriel R.]

what is your insert size?