All chromosomes for mapping?

Question

[caggtaagtat]

Hi,

I'm still getting to know STAR and wanted to ask you, if you just use Chromosomes 1 to 22 and X,Y, MT or also all the other chromosomes with longer names in your fasta reference file for mapping human reads?

[GenoMax]

It depends on what your aim is. If you care about those (I assume you are referring to random/haplotype type entries) then you will have to. This may be of interest.

[caggtaagtat]

Thanks, I'm interested in alternative splicing in human cells.

Answer 1

[Devon Ryan]

Include the various unplaced scaffolds but not the haplotype alleles. The latter will completely screw things up. Including the former will slightly decrease false-positive alignments.

[caggtaagtat]

Alright, thank you. So since I'm looking at alternative splicing, including the non-chromosomal fasta file in my genome could may be rewarding than. Do you know where the unplaced scaffolds come from?

[Devon Ryan]

The random contigs fall into two groups: those with a known chromosome of origin and those with no known chromosome of origin. In the latter case I presume these actually have multiple copies, though I've never checked. In the former case it's likely that these are regions that just to integrate into the assembly well.

[caggtaagtat]

Ok thanks, I will included the fasta files (from ensembl) of chromosome 1-22, X, Y, of mitochondrial and non-chromosomal DNA for my mapping with STAR from now on.