Negative FREELK1 - FREELK0 for --freemix

[raqs...@gmail.com]

Hi,

I have been getting negative values (all kinds of range) for FREELK1 - FREELK0 independently of the FREEMIX value.

Command line used is 

verifyBamID --vcf $1 --bam $2 --out $3 --verbose --ignoreRG --site --minQ 20

Two example outputs from different BAMs and VCFs:

 #SEQ_ID RG CHIP_ID #SNPS #READS AVG_DP FREEMIX FREELK1 FREELK0 FREE_RH FREE_RA CHIPMIX CHIPLK1 CHIPLK0 CHIP_RH CHIP_RA DPREF RDPHET RDPALT

SI_2 ALL NA 11341430 9579343 0.84 0.49992 3276580.38 3362001.50 NA NA NA NA NA NA NA NA NA NA

or

#SEQ_ID RG CHIP_ID #SNPS #READS AVG_DP FREEMIX FREELK1 FREELK0 FREE_RH FREE_RA CHIPMIX CHIPLK1 CHIPLK0 CHIP_RH CHIP_RA DPREF RDPHET RDPALT

79_C4_CD4naive ALL NA 2615527 438963 0.17 0.00468 171452.74 171803.39 NA NA NA NA NA NA NA NA NA NA

Best regards,

Raquel

[Hyun Min Kang]

You need to negate the likelihood value. (it is negative log-likelihood)

Hyun.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.
To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[raqs...@gmail.com]

Thank you Hyun!

So if [FREEMIX] >= 0.03 and absolute(FREELK1 - FREELK0) is something like 100 we could say is possible that the sample is contaminated with other sample?

[Hyun Min Kang]

Yes, I think that is a reasonable threshold. Obviously, the contamination wouldn't matter much if the estimated contamination rate is really small. 

Thanks,

Hyun.