microbiomedata / kaiko_metaproteome Goto Github PK

Add more detail to the output and intermediate files of kaiko. Number of proteins per taxa, taxa dependency.

This will allow us to consider more complex logic when building the FASTA file.

Currently the taxa -> FASTA logic is rough around the edges.

Main goal is lower FASTA size, while keeping accurate taxa identification. Taxa identification looks good, but the FASTAs are huge.

Create FastAPI functions to help with deployment of Kaiko into existing workflows.

Details to follow as we get closer to finishing Kaiko, add them here.

The current model is from ~2017. Current transformer based models may give us huge improvements in denovo identification.

This would be a more open ended path. Thankfully, we can use denovo sequencing model within Kaiko.

Use X-Tandem quick search to test how well raw spectra protein search performs with a particular taxa.