-
Hello, Thanks so much for publishing your awesome models. I was curious what the easiest way would be to score a mutant vs a wt sequence with your zero shot ESM1v method. And then compare that to fine-tuning the model with a specific MSA and performing the same task? Thanks so much! |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
Hi Franziska, We provide example code for scoring a mutant vs. a wildtype sequence using the zero-shot ESM-1v method here: https://github.com/facebookresearch/esm/blob/master/variant-prediction/predict.py The script is setup to accept a wildtype sequence ( In regards to fine-tuning the model on an MSA, please refer to the general discussion here: #33 Hope that helps! |
Beta Was this translation helpful? Give feedback.
Hi Franziska,
We provide example code for scoring a mutant vs. a wildtype sequence using the zero-shot ESM-1v method here: https://github.com/facebookresearch/esm/blob/master/variant-prediction/predict.py
The script is setup to accept a wildtype sequence (
"--sequence"
) and a CSV file with the mutations over which you'd like to predict ("--dms-input"
). If you have just a single mutant sequence, you can either modify the script or use a simple CSV file that contains a single row.In regards to fine-tuning the model on an MSA, please refer to the general discussion here: #33
Hope that helps!
Joshua