search

zhou-lab / biscuit

BISulfite-seq CUI Toolkit
Other
62 stars 24 forks source link

Another SNP calling error #9

Open jdidion opened 7 years ago

jdidion commented 7 years ago

"a or b too big, or MAXIT too small in betacf: Numerical result out of range"

Looks like a place where there just needs to be some exception handling.

jdidion commented 7 years ago

This looks like it might be a read depth issue. I've successfully called SNPs on all autosomes except two specific regions: chr1:121400000-121500000 and chr17:22000000-22100000 (GRCh37). The first is mostly satellite DNA, the second harbors a TRNA. This tells me that I need to exclude from SNP calling regions that are outliers for coverage. It would be convenient to have a max read depth parameter (i.e. the opposite of -k).

jdidion commented 7 years ago

I narrowed the second interval down to chr17:22020600-22020900, which seems to contain a sequence very similar to the mitochondria.

zwdzwd commented 4 days ago

Increasing MAXIT in stats.h temporarily fix the error

#define MAXIT 10000 // previously 100

will try to find a better solution when I get a chance