Fatemeh Zare‐Mirakabad

This paper predicts protein sequence profiles using a ProtAlbert transformer without alignment, showing high similarity to HSSP profiles.

The paper evaluates four frameworks using similarity vectors of drug and adverse reaction features to predict adverse drug reactions, finding that frameworks with fewer features achieve higher accuracy.

This paper evaluates sequence representations for improving drug-target binding affinity prediction without requiring protein structures.

This paper extends Wang's protein design model by incorporating the Blosum62 substitution matrix to account for amino acid replacement probabilities during sequence design.