--keepDupes and -F 0x400 gives me different results

[crazyhottommy]

Hi,

I have some single end RRBS-seq data. I aligned with bwa-meth and marked the duplicates with samblaster.

one expects to see a lot of duplicated read for RRBS data, because the enzyme cutting site is the same.

MethylDackel (0.3.0-3-g084d926) gives me quite different results specifying --keepDupes or -F 0x400. Are these two options the same?

thanks,
Tommy

[dpryan79]

-F 0x400 is exactly the opposite of --keepDupes. -F in MethylDackel is the same as -F in samtools view, so you're then instructing it to ignore duplicates (but then include secondary and supplemental alignments, since you've overridden the default value of 0xF00 or 3840).

[crazyhottommy]

Thanks. so, If I need to include the duplicates, I can do:

MethylDackel --keepDupes

# or

MethylDackel -F 0x100 -F 0x200 -F 0x800

what if one uses both --keepDupes and -F together? and the order of the flag also matters seems.

e.g. This seems to give different results.

MethylDackel -F 0x400  --keepDupes
MethylDackel   --keepDupes -F 0x400

Thanks,
Tommy

[dpryan79]

The -F option is processed first when reads are filtered, so if you filter it then --keepDupes is ignored. This looks like a bug, actually, since I'm now not sure that --keepDupes is properly doing anything since it isn't changing the -F default. I think I should change the -F default to 0xB00, I'll have to test that.

[crazyhottommy]

Thanks for the detail. I just tested

MethylDackel -F 0x100 -F 0x200 -F0x800
MethylDackel   --keepDupes  -F 0x100 -F 0x200 -F0x800

I thought these two are the same. -F 0x100 -F 0x200 -F0x800 will keep the duplicates, but one needs to specify --keepDupes as well for the counting.

For my purpose, I will need to do to keep the duplicates but discarding failed qc, secondary and supplementary reads:

MethylDackel   --keepDupes  -F 0x100 -F 0x200 -F0x800

right?

Thanks,
Tommy

[dpryan79]

-F 0xB00 rather than multiple -F values, but otherwise yes. I'll make that the default in the next bug fix version.

[crazyhottommy]

great! thanks for the explaining and user support!

[dpryan79]

Thanks for reporting the bug!

[dpryan79]

Sorry for the delay on this, it's now fixed in the 0.4.0 release.