BAM file size difference between 2 sequencing platforms

[Ewiwi]

Hello everyone,

It is my first time posting so I don't really know if I'm at the right place.
I am new to sequencing and still a student so please tell me if my question is a dumb question:

I'm working on a project in which we have sequencing datas of people coming from 2 platforms: one from Thermofischer and one from Illumina. However, the BAM file size per individual is 8 times bigger for the Thermofischer originated datas compared to the Illumina datas. How can that be explained ? Can it be the sequencing method (Sanger/ Ion torrent for Thermofischer), the coverage, or the alignment software used ? (For Illumina dataset, the iSAAC pipeline was used apparently).

I really want to get a better understanding of all these things so if you have advices or documents that might help me, I'll be happy to take it.

Thank you very much!!!

[Brian Bushnell]

I'm working on a project in which we have sequencing datas of people coming from 2 platforms: one from Thermofischer and one from Illumina. However, the BAM file size per individual is 8 times bigger for the Thermofischer originated datas compared to the Illumina datas. How can that be explained ? Can it be the sequencing method (Sanger/ Ion torrent for Thermofischer), the coverage, or the alignment software used ?

Yes, it could be any of those! Normally a bam file's size is most closely related to the coverage, but the contents can vary considerably based on all of those factors - some aligners produce multiple lines per read, and add a bunch of optional fields to each line. The bam generation settings also have a large impact - the file could be compressed or uncompressed, sorted or unsorted. And different platforms produce very different quality score profiles; much of the size of a compressed bam comes from the quality scores.

[Ewiwi]

Thank you very much for your reply. I start to understand better how are generated BAM files now. So I guess that these different datasets can't really be compared in downstream applications even if they were aligned on the same reference genome? Since the alignment methods may differ (and I'm almost sure they are) can I really compare them in the end ?

Thank you for helping a poor student

[Brian Bushnell]

You can usually compare data qualitatively if it was aligned by the same version of the same aligner, using the same reference.

You cannot usefully compare data quantitatively if:

1) The aligner was different.
2) The reference was different.
3) The platform was different.
4) The preprocessing was different.
5) The lysing was different.
6) The amplification was different.
7) The DNA quantity or concentration was different.
8) Anything was different, other than the sample.

You can usefully compare things if a single variable differs, for the purposes of evaluating the single variable's effect on results. It is pointless to compare things when there are many differences in methodology, unless you are following something like a rigorous fractional factorial design, which requires design prior to experimentation.

[Ewiwi]

Thank you very much for your quick reply. I'll keep that in mind !