Hello
I'm reading a paper and in the experimental procedures there is a part about variant filtrations. They used filter which exclude variants that were (1) labeled as VQLOW in all individuals, (2) clustered in >10% of individuals from either cohort, (3) variants missing in >10% of either cohort, (4) with median coverage 100 in either cohort, (5) in simple repeats, homopolymer repeats >6bp, segmental duplications, microsatellite repeats, or low-complexity repeats, (6) out of the Hardy-Weinberg Equilibrium and (7) in non-unique 36-mers.
I don't exactly understand criterion (2), (4) and (7). Why is it a problem when the variants are clustered? Why is a coverage of more than 100 a problem? And to what are these 36-mers referring? And has (5) something to do with the fact that it is difficult to map repeat regions?
Hi Eline,
My guess would be:
(4) low median coverage - not relaibly sequenced; very high median coverage - part of repetitive sequence, hence non-reliable mapping and variant calling
(5) difficult to map, reference might be inadequate. I sequenced once an inbred mouse strain and after mapping, regions like olfactory receptors appeared heterozygous.
(2) clustered variants - this I am less sure; I do not know how much "clustered" they are; if very close, might suggest errors in PCR/mapping maybe?
(7) what was the read length in the sequencing? Maybe again this is a mapping issue?
It would help if you would provide some link to the paper.
- Badges
- Science topic
- Similar topics
- Correction