none correlation between modification position and +/- 1

[DelphIONe]

Hi Marcus,

Your tool is the only one that allows you to do an analysis per read, it's very useful! Thanks for this. I have a sequence with a modification detected by the model_sample_compare method (and the parameters --fishers-method-context 0 --sample-only-estimates). On the plot from tombo plot per_read (with --genome-locations --num-bases 301 --num-reads 200 --box-center parameter), I observe no correlation in terms of color (-log10(pval)) between one position of a read and the next or the previous one. Can you explain this ? The modification must have an impact on the kmer and not just on a base. To convince myself I used your API (thank you very much, it's very useful!) to obtain the pvalue per reads per position and I don't see any correlation. I don't understand this, if you could explain it to me I'd be very grateful.

[marcus1487]

I would suggest that you move your analyses over to Remora which can also perform per-read signal level analyses from standard nanopore data processing (POD5 and BAM). See the notebooks showing this functionality can be found here.

For your particular issue I am not quite sure what you mean. The per-read metrics inspects metrics on individual reads (and can be quite noisy), while the model_sample_compare method compares groups of signals, and thus may provide much more power for detecting shifts in signal metrics. It would be much easier to debug any issues with examples similar to the notebooks from Remora though, so shifting to this paradigm would be highly recommended.

[DelphIONe]

Thanks for your reply and sorry for my delay. I'm intersting by remora and I would like follow your advice but Remora can work by "sample compare" method as Tombo ? The doc is not very explicit on this subjet. Thanks for your help.