Confused with multiple runs per SRA entry

Question

[Maycon]

I'm trying to process the raw data from this CHIP-seq project to eventually perform a peak-calling analysis (basically reproduce the methodology described in the paper) and eventually a differential peak analysis to verify if there are differences in H3K9ac binding between the treatment and control groups for a particular gene. I'm confused, though, because some samples (like this one or even input samples) have two runs while others have only one (like most the H3K9me2 samples and one of the input samples). There's no description in the Methodology for why that is, at least not that I've seen. The metadata for each run is also not very clarifying. I don't believe they are supposed to be biological replicates since those are divided into different SRX accessions, properly named (e.g. the paper describes there are two replicates for both treatment and control for H3K9ac, and there are 4 different GEO samples for those, appropriately named). How can I tell if those multiple runs per SRX are technical replicates and not biological replicates? And if they are technical replicates, can I just cat the FASTQ files prior to aligning with bwa?

Answer 1

[GenoMax]

If you look at SRA run selector entry for this experiment perhaps things will become clear.

SRA accessions with the same sample name appear to be technical sequencing replicates (they could be two separate libraries made from the sample but people rarely do that and that does not seem to be the case based on what you describe above). Where as the biological replicates are so designated e.g SAMN03377053 and SAMN03377054 are biological replicates. You can cat the technical sequencing replicates when you do the alignments.