Visualising f5c resquiggle output in UNCALLED4

[hasindu2008]

Would it be possible to support f5c resquiggle output in UNCALLED4 visualisation? It would be quite useful for us to compare Guppy movetable-based alignment against the resquiggle?

[skovaka]

This should be possible via the uncalled convert command. Currently the only way to compare with the projected guppy alignments is to additionally run uncalled dtw and write to the same SQL file. I should add the standalone basecaller alignment command, but for now the series of commands would be something like this:

uncalled covert --eventalign-in <file.tsv> ... --sql-out <alns.db>:f5c #f5c is the name of the f5c "track"
uncalled dtw ... --sql-out <alns.db>:unc #compute guppy alignment and perform DTW
uncalled dotplot --sql-in alns.db #visualize f5c, uncalled, and basecaller alignments

This is more work than it should be, and soon I'll push an update which will allow you directly visualize an eventalign file. Stay tuned!

[hasindu2008]

I think I was unclear in my question. Sorry about that. Some time ago I managed to get f5c eventalign results to correctly be loaded. This is about f5c resquiggle. It is a recent module I added to f5c, to output alignment of the signal to the base called read (where as, eventalign outputs the alignment of signal to reference). The output format is a TSV, but slightly different: https://hasindu2008.github.io/f5c/docs/output

[skovaka]

Ah yes, sorry I forgot resquiggle refered to aligning to the basecalled reads. That will take some reworking to handle reads not aligning to a common reference, but it does sound useful. I will work on it and let you know when I make progress

[hasindu2008]

Cool. Also, now are you supporting BAM move tables - is there an example of the steps for that?

[skovaka]

I do support BAM move tables now. You just need to run Guppy with the "--moves_out" and "--align_ref" options, merge the resulting BAM files with samtools, then input the merged bam file to "--bam-in" in the uncalled dtw command. There is one known issue in that Guppy only outputs moves in primary alignments, so you won't be able to align secondary alignments with this method currently. I will work on a workaround for that

[hasindu2008]

Primaries are enough for me. But loading takes so much time, are you still accessing the signal data from FAST5 when using this BAM method? Looks like most of the time is spent on IO (this is a promethION dataset on an NFS storage).