marcpaga / sturgeon

Sturgeon is a CNS neural network classifier
https://www.nature.com/articles/s41586-023-06615-2
Other
32 stars 8 forks source link

Update Example usage with demo data (megalodon txt files) #10

Closed chinmaysharmacs10 closed 8 months ago

chinmaysharmacs10 commented 9 months ago

The commands in the read me to run live mode prediction on megalodon txt files do not mention path for probes file. As a result, the code attempts to use the default probes.bed file, which doesn't exist in '/venv/lib/python3.9/site-packages/sturgeon/include/static/' resulting in an error.

I would suggest to update the run command to include the T2T probes file:

''' sturgeon live \ -i demo/mega \ -o demo/mega/out_live \ -s megalodon \ --model-files PATH_TO_MODEL_DIR/general.zip \ --probes-file ./venv/lib/python3.9/site-packages/sturgeon/include/static/probes_chm13v2.bed \ --plot-results '''

marcpaga commented 8 months ago

This is a problem due to the addition of probe bed files for chm13 and hg38. This should fix the bug https://github.com/marcpaga/sturgeon/commit/bf2f12211b6305930dc006318b7540bc5dad62f7. I have added the option to pass the desired reference genome as with predict https://github.com/marcpaga/sturgeon/commit/67fb704356e3992c71898c468f12b9fdc488633f

chinmaysharmacs10 commented 8 months ago

Hey @marcpaga,

In your last comment you mentioned "I have added the option to pass the desired reference genome as with predict".

However, the predict mode doesn't have a reference genome argument since it takes bed files as input, and bed files are already aligned with the reference genome. Is that the correct understanding?

marcpaga commented 8 months ago

Yes, sorry. I meant the inputtobed utility.