How doing a prediction if number of columns doesn't match?

Question

[Francesco]

Hi! i'm a beginner in ML world and i would to calculate some proteins descriptors using MathFeature and iFeature (python libraries). Some of these descriptors generate a number of columns equal to the max protein lenght. When generating the model, if i try to predict a list of shorter proteins, the number of columns are different and i get error. how can i resolve this issue? thank you

Answer 1

[Mensur Dlakic]

Average representations from protein language models (pLMs) will give you vectors of the same size regardless of protein length. There are many pLMs, only listing the ones I tried.

• https://github.com/agemagician/ProtTrans
• https://github.com/facebookresearch/esm