Questions about strands

[bschilder]

@Zhihan-Leo-Liu had some questions about how GVL handles strand information.

I believe this also came up with @Al-Murphy recently, and I couldn't really answer it myself.

The GVL docs mention this about strands, but I don't really understand what it means.

How can I get multiple tracks/stranded data?
If you provide multiple tracks to gvl.write(), all of them can be returned simultaneously from the resulting Dataset and placed along the track axis, sorted by name. By default, a Dataset sets all tracks to active when opened. i.e. tracks have shape (batch, tracks, [ploidy], length).

In my own tests in the splicing functionality, I noticed that when I take the reverse complement of transcripts that fall on the negative strand ("-"), using the reverse_complement function in Biopython, the number of stop codons per sequence decreases (which is generally a good sign).

@Zhihan-Leo-Liu @Al-Murphy please direct any more specific questions you have to @d-laub here or via Slack.

[d-laub]

Good question! I can clarify this in the docs. By default, GVL Datasets have their rc_neg setting set to True. This means any regions in the dataset that are on the negative strand will automatically be reverse complemented (or just reversed for tracks). To clarify, stranded tracks would be when we counted reads that aligned to either the positive or negative strand and put them into separate BigWigs. For example, Adam He from the Danko Lab had stranded tracks from PRO-cap sequencing that I believe is similar (or perhaps the same) to what @Al-Murphy is using.

For the splicing, is that behavior persisting after you've adjusted the feature coordinates to be 0-based?

Let me know if I can clarify this more, it's definitely a bit confusing because of the jargon/identical naming.

[bschilder]

Good question! I can clarify this in the docs. By default, GVL Datasets have their rc_neg setting set to True. This means any regions in the dataset that are on the negative strand will automatically be reverse complemented (or just reversed for tracks). To clarify, stranded tracks would be when we counted reads that aligned to either the positive or negative strand and put them into separate BigWigs. For example, Adam He from the Danko Lab had stranded tracks from PRO-cap sequencing that I believe is similar (or perhaps the same) to what @Al-Murphy is using.

For the splicing, is that behavior persisting after you've adjusted the feature coordinates to be 0-based?

Let me know if I can clarify this more, it's definitely a bit confusing because of the jargon/identical naming.

Ahh I see. This is sort of what I remembered from previous convos, but wasn't sure. So I shouldn't actually be taking the reverse complement myself bc it would be doing it twice.

Yeah, I'm still having some issues getting decent seq sim between GVL and Haplosaurus even with the your suggestions. But will send some updates on that Issue separately.

[d-laub]

Is this still an open question? Otherwise I'll close this EOD tomorrow.

[bschilder]

@Zhihan-Leo-Liu