search

RahmanTeam / DECoN

50 stars 29 forks source link

samples off by one #40

Open atimms opened 2 years ago

atimms commented 2 years ago

Using the latest version of decon I'm getting a consistent error. The first bam file is always described in the Failures.txt file as: All Whole sample All Low correlation: 0.29554106282364, Low median read depth (FPKM): 0.405555555555556 In addition the CNV calls are always off by one so known CNVs are assigned to the incorrect sample. Any help would be appreciated. Andrew

atimms commented 2 years ago

this looks like an issue with the counts/ExomeCounts where the GC column is being assigned to the first sample and then all the samples are off by one. Has anyone seen this or know of a resolution?

fishcakess commented 1 year ago

Has anyone seen this or know of a resolution?

Andrew change this part in the Rscript IdentifyFailures.R

colnames(ExomeCount)[1:length(sample.names)+4]=sample.names

Change that 4 to a 5 then save it.

wcarre commented 1 year ago

same issue in makeCNVcalls.R. Ligne 69: colnames(ExomeCount)[1:length(sample.names)+4]=sample.names it should be colnames(ExomeCount)[1:length(sample.names)+5]=sample.names

Otherwise all the samples / CNV are offsets.