Closed will-NYGC closed 3 years ago
Sorry for the slow response. The BAM that pore-c creates appends information to the read name to make it work with WhatsHap. It also contains alignments that fail the pore-c filters and so probably shouldn't be used for your methylation analysis. I recently pushed a new version of the tools (0.4.0) with a tool to prep your BAM for methylation calling: https://nanoporetech.github.io/pore-c/cli.html#filter-bam
There's also a new version of the Pore-C-Snakemake pipeline that does the methylation calling for you.
Thanks, I did manage to figure this out and formatted the BAMs to make it work. I see the new snakemake uses f5c to make the meth calls per batch. this is great! will use this new pipeline moving forward.
Glad to hear that - sorry again for the delay in responding.
I am trying to call methylation on the guppy called fastq's and Pore-C Snakemake BAM outputs with nanopolish. I am using a test fastq with 50000 reads. All 50000 are indexed properly but when I use nanopolish call-methylation I get lots of "bad fast5" and no calls.
I don't quite understand why the call-methylation sees 141242 total reads and why 140191 have "bad fast5". When I look at the readdb index file, all fast5 locations are readable.
Does the porec aligned BAM split the concatemers into separate reads? Is that why the BAM contains 235538 reads even though the fastq only has 50000.