Running predict.py on sample fasta file

[aparsan]

Hi! I'm new to deepgo and Linux and am attempting to run deepgo2 on my own fasta file, but am getting a time out error.

---

python predict.py -if example.fasta -d cuda

I was wondering how long it should take to make one prediction? I'm running on 1 a-100-GPU and with a memory of 70 GBs. is there's anything I'm doing wrong in terms of how the path is presented/where the example file should be relative to the predict.py file. Thanks so much, would appreciate any help!

[coolmaksat]

Hi, it should be quite fast on a A-100 GPU. But, when you run it first time it downloads ESM2 model which may take some time depending on your internet. Can you provide the output?

[aparsan]

I was able to get it to work, thank you so much!

Another quick question – I'd like to train and test the model on my own data. Is there a format/file structure that would allow me to do so most fluidly and with the least amount of modifications to train.py?

[coolmaksat]

Hi, Please have a look at gendata/deepgo2_data.py and gendata/uni2pandas.py scripts that I use to generate training/validation/testing splits.