Open carlosmag opened 4 years ago
Thanks for the suggestion. I agree these bias information is also important for downstream analysis. I'll design data structures to keep track of these information. The strand bias means an allele only occurs at specific strand, while cycle bias means an allele occurs at the beginning sequencing cycle. However, I don't understand what the placement bias means. Could you please explain it for me? Thank you very much. Moreover, Erik Garrison also mentioned allele imbalance. What does it mean? It seems like we can only observe an allele in a small portion of read alignments covering the locus. Is it correct?
Thanks for considering bias info! Read placement bias info is used to filter for sites with reads balanced to each side (left and right), as explained in freebayes in depth: model, filtering, and walkthrough, pages 16-17, Erik Garrison 2015. I am not familiar with allele imbalance, but it seems to be related with this:
from Galaxy Training!, Calling variants in non-diploid systems, Anton Nekrutenko 2018
it reflects significant deviation from the diploid (50/50) expectation (see https://galaxyproject.github.io/training-material/topics/variant-analysis/images/freebayes.pdf for more details)
Moreover, Erik Garrison also mentioned allele imbalance. What does it mean? It seems like we can only observe an allele in a small portion of read alignments covering the locus. Is it correct?
@ekg - can you confirm the definition?
At a heterozygous genotype the fraction of observations for the pair of alleles is unlikely under an unbiased binomial sampling model.
On Tue, Oct 8, 2019, 08:51 Torsten Seemann notifications@github.com wrote:
Moreover, Erik Garrison also mentioned allele imbalance. What does it mean? It seems like we can only observe an allele in a small portion of read alignments covering the locus. Is it correct?
@ekg https://github.com/ekg - can you confirm the definition?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/hsinnan75/MapCaller/issues/25?email_source=notifications&email_token=AABDQEMF2PHGS3D2HCAXU73QNQUYTA5CNFSM4I52KXU2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEATBSZA#issuecomment-539367780, or mute the thread https://github.com/notifications/unsubscribe-auth/AABDQEPFAYRHG7QPLMDH5TTQNQUYTANCNFSM4I52KXUQ .
Hi!
FreeBayes reports 4 INFO fields that are used to assess strand and read placement bias.
Graphical illustration:
adapted from freebayes in depth: model, filtering, and walkthrough, page 9, Erik Garrison 2015*
Can you include these metrics in MapCaller? Thank you!