thermostability prediction results for esmfold versus colabfold plddt differences

[blazespinnaker]

Currently there is a thermostability prediction competition on Kaggle. One of the features we're looking at is using plddt difference. On the leaderboard, using colabfold models it gets about .297, but is still useful as it correlates poorly with other features. Using esmfold, we're only seeing about 0.091 on the leaderboard.

You can see the notebooks here:

Using esm dataset generated from web api:
https://www.kaggle.com/code/kaggleqrdl/chris-diff-features-with-ropeonmars-esm

Using colabfold generated dataset from a participant:
https://www.kaggle.com/code/cdeotte/difference-features-lb-0-600