I've been experimenting with Majiq Build v2.2-e25c4ac to determine if I understand how to use it.
I have a test dataset from two mouse strains, one of which has a knockout of two exons in a gene. We have 9 RNASeq samples for each genotype, with a sequencing depth of 40-50m, paired end, 150bp reads per sample, mapped to the mm10 ensemble genome and ensemble v102 annotation with STAR.
MAJIQ does find a few (~10-20) very minor (hard to spot visually) LSV between the two strains, but they are only located in non-standard chromosomes like chrX_random, chrUn_JH584304 etc. Does this indicate something is off? What is a normal range of numbers of LSV you would expect to find in an actual animal experiment? 100s? 1000s?
I am wondering if MAJIQ is ignoring the actual standard chromosomes for some reason? In particular, it never identifies the actual exon knockout we engineered as a variant.
I can see the variant very clearly myself in the bam files, but it just isn't being spotted computationally. Do you think I am doing something wrong or is this something the program normally wouldn't detect for some reason?