search

zhanxw / rvtests

Rare variant test software for next generation sequencing data
131 stars 41 forks source link

Allele frequency column does not match N_REF,N_HET,N_ALT column #52

Open sgrosche opened 6 years ago

sgrosche commented 6 years ago

Dear Xiaowei, thank you for developing rvtests. I am using your tool and have realized that from version 20160504 on the allele frequencies reported and the INFORMATIVE_ALT_AC column do not match the numbers reported in the N_REF,N_HET and N_ALT columns if the dosage tag (--dosage) is used. Also the results in the allele frequency column, the INFORMATIVE_ALT_AC column and the ALT_EFFSIZE column differ between version 20151007 and newer versions when using the --dosage tag and the same input file. The N_REF,N_HET and N_ALT columns, pvalues and the U_STAT,SQRT_V_STAT columns remain the same between versions. Could you tell me where the differences occur from? Thank you very much and best wishes, Sarah

zhanxw commented 5 years ago

It is most likely that dosages are not integers, and there are rounding errors. For example, if the dosage is 1.5, and the genotype is 1, you will get allele counts of 2 and 1 respectively.