@Daniel Lundin : I've published a release of the EPA-ng version of SATIVA, tagged

v0.9.3.3

, and opened a bioconda PR for it (package

sativa-epang

). SATIVA can now be called step by step: (i) build the reference, (ii) prepare the folds, (iii) place, (iv) score. Each step can be its own Nextflow process now. Each fold directory holds everything needed to place it (the tree, the alignment, the queries, the model) along with the EPA-ng command to run in it. So the placement step needs neither the full reference, nor the taxonomy, nor even SATIVA installed. Run with no options, SATIVA still does everything on its own, as before. The code is here: github.com/Aaramis/sativa-epang :

CHANGES-epa-ng.md

documents the steps, the fold layout and the manifest.

Fantastic! cattyping I'll go ahead on my side

Thanks Daniel, i will write a message on it !

As I've been planning to run RAXML-NG instead of RAXML for the reference tree, Claude has a question for you: "does -r accept a .refjson/.model pair built by something other than his own -stage reference (e.g. one we construct from a RAxML-NG run), or is refjson intrinsically tied to how RAxML's reference step serializes things? The doc only documents reusing a reference sativa-epang itself built earlier. If refjson turns out to be RAxML-specific, we may be stuck either running his RAxML reference step just to get a compatible refjson (and eating the model-testing loss), or reverse-engineering the format ourselves. That answer decides whether RAxML-NG and the sativa-epang staged pipeline are actually compatible at all."?

No, the refjson isn't tied to RAxML's serialisation. The only genuinely RAxML-8-specific field is

binary_model

, and it's inert in the EPA-ng path, I blanked it and the output came out identical. What's actually needed is the tree with a

B=

id per branch, plus the taxonomy, sequences and taxonomic code. Those ids are just identifiers, and

branch_tax_map

is derived from tree + taxonomy in pure Python. Models are fine too EPA-ng's

-m

takes a RAxML-NG model string or file directly. But if you want, I can fix that quickly so it accepts a ready-made tree and model instead of running RAxML itself, your RAxML-NG output would then plug straight in. Want that?

Yes please, that would fit my pipeline perfectly.

Done :

-reftree

and

-refmodel

are in, tag

v0.9.3.4

. Point them at your RAxML-NG tree and model and the reference step skips RAxML entirely. Feeding a reference's own tree back through them reproduces its output byte for byte on my tests

Will this invalidate your bioconda recipe request?

I don't think, same PR, I just bumped it to

v0.9.3.4

. Nobody had reviewed it yet, and the review label is still on, so nothing was lost; CI is re-running on the new version.

I hope I don't! 🙂