Pipeline revisions - kraken, flye, etc

[tavareshugo]

Update the pipeline to use de-novo assembled genomes:

• flye, possibly downsampling first with trycycler subsample (see perfect genome assembly)
• medaka consensus for polishing the assembly
• prokka for annotating genes (if we have time we can also add how to use bacta)

Lower priority is to use kraken2 instead of mash.

• Kraken database is here, we can use "Standard-8"

[tavareshugo]

We have a script that runs a loop through several samples, based on a samplesheet.

Note about medaka_consensus: if a GPU is detected, then tensorflow will use it automatically. But because our laptop GPUs are relatively low-spec (4GB on our NVIDIA cards), the default medaka settings make the pipeline fail. There are two possibilities:

• -b 10 --> set this option with medaka_consensus (source)
• export CUDA_VISIBLE_DEVICES="" --> to turn off GPU usage (source)

[tavareshugo]

We now have a custom-made shell script that does all these steps.

However, some notes of alternative pipelines to consider in the future:

• https://bactopia.github.io/v2.2.0/