search

al-mcintyre / mCaller

A python program to call methylation (m6A in DNA) from nanopore signal data
MIT License
43 stars 16 forks source link

nanopolish extract non-functional #24

Closed wshropshire closed 3 years ago

wshropshire commented 3 years ago

Hello,

I just wanted to let you know that nanopolish extract cannot extract fast5 file pathway information to incorporate into read headers. The authors of the program have basically noted its obsolete for current versions of nanopolish. I have a set of four isolates where two are complementary mutants that have a methylase knocked out. I first de-multiplexed fast5 previously with Deepbinner since I was concerned about the clonal nature of these isolates. However, after thinking about the pathway more clearly, I'm thinking besides adding more computational time, omitting demultiplexing shouldn't be a concern for downstream results. Am I correct in thinking that demultiplexing fast5 files is not necessary for downstream analysis?

Thanks!

Will

al-mcintyre commented 3 years ago

Hi Will,

What's your experimental design and what's the goal of your downstream analysis? You have one run with four barcoded samples? Do you want to identify differences between the mutants and wildtype isolates? Differences between the two complementary mutants? Or methylated motifs in common among all of the isolates?

After running mCaller, you will have probabilities of methylation listed for each site of interest per read. I assume at some point you will need to know with which condition each read is associated for your analysis, but there's no issue with demultiplexing after starting, depending on what is most convenient.

Has the update to nanopolish extract been an issue?

wshropshire commented 3 years ago

I have two sets of isolates (6180 and 1057) where one is WT and the other has a knocked out methylase gene from a type I R-M system. We have PacBio data for one set of isolates (6180) and predicted motifs for the other (1057).

I have already demultiplexed my samples though and using the barcoded reads for mCaller, therefore I don't think it will be an issue that the fast5 files weren't demultiplexed initially. I don't believe that nanopolish extract though is currently functional.

al-mcintyre commented 3 years ago

Ok, did the instructions using nanopolish index work instead? I can take extract out of the README to avoid confusion, since it's been a few years now.

wshropshire commented 3 years ago

Tentatively I can say yes, it appears nanopolish index works, but let me confirm once I am able to parse through the mCaller.py output. I believe though per the authors of nanopolish that they are deprecating the nanopolish extract command in favor indexing instead.

Best,

Will

al-mcintyre commented 3 years ago

Closing for now, assuming indexing is fine. I've removed the old reference to extract from the README.