Question regarding the pipeline

[beerchen007]

Hello!
I have a question: Is it possible to join the pipeline later? I already have alignments and would like to use them to create a ML-Tree and possibly more tests. So I don't need the first steps of the pipeline.

Since “phylociraptor” creates checkpoints and yaml files, I don't know if this is possible. What is your opinion?

Thank you in advance!
Best,
Vera

[onestardao]

this is a classic case of ProblemMap No.6 — pipeline bootstrap ordering.
trying to “join later” in the flow sounds simple (skip alignment → start from tree building), but because the pipeline state is encoded in checkpoints and yaml configs, the order dependency is hard-baked. skipping ahead often means the later stages can’t find the expected artifacts or metadata.

we’ve catalogued 16 such pipeline / RAG failure modes with reproducible fixes. if you’d like the full list (with corresponding repair strategies), let me know — it’s easier than trial-and-error with partial restarts.