Duplicated and missing entries in .methphased.vcf about methphaser HOT 6 OPEN

I will download mm39 and take a look, thanks!

from methphaser.

Should be a bug, could you please show me the vcf file? I have never used MethPhaser on mice samples so a little example could help.

from methphaser.

It's not public data - I'll have to ask if it could be shared. I may get back to you on this at a later date. Sorry I can't be more helpful now.

Although one thing I encountered from the current implementation is that, since I filtered the .vcf file for only those on chromosome 1-19 (mouse autosomes), phase_block_df (defined here) defaults the dtype for the chr column to be int instead of str, and the subsequent comparison fails as int(chromosome) == str(chromosome) returns False. As a result all chromosomes were being skipped. Adding dtype={'chr' : 'str' } to the pd.read_csv() call in defining phase_block_df fixed the issue. A similar issue is also in methphasing.

from methphaser.

I think that was because most human autosomes can be parsed as int but I can fix that, thanks for pointing out! Btw what is your reference genome headers like? Maybe that was different from human autosomes and could be the issue, not sure.

from methphaser.

We use mm39 as reference.

Now I think about it - this line here reads in info from the reference (and the next line gets the chromosome column). If there are non-autosome entries in the reference, there will be non-numerics in their names, and the chromosome column dtype will be str. In this case, if the .vcf file is filtered for autosomes and the chromosome column dtype gets defaulted to int, the int to str comparison fails.

from methphaser.

Yeah mm39 starts with something like NC_000067.7 and HG38 starts with chr22. I will modify the dataframe reader, thanks for pointing that out

from methphaser.