You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The problem: In #14, @bmreilly observed that adding more samples resulted in fewer sites tested for DM.
The reason:060381b moved the minCount/maxCount filter from methylSigReadDataSingleFile() to methylSigReadData() after destranding. This way destranding would rescue +/- sites that individually were < minCount, but when destranded were > minCount. This fix introduced another bug whereby a site in any sample failing minCount/maxCount would remove this site from every sample. This behavior is consistent with the behavior in #14.
The fix: This release refactors methylSigReadData() and methylSigReadDataSingleFile() so that destranding and the minCount/maxCount filter are in methylSigReadDataSingleFile(), and only the site in the offending sample will be minCount/maxCount filtered. Included in the fix are new unit tests and new test files to fully ensure the behavior we expect.
Accessor for findOverlaps()
9ba232c accesses queryHits with accessor function queryHits() instead of slots @ when filtering SNPs. Accessing with @ was causing Travis-CI build failures.