Thomas Lemmin

H3BERTa is a transformer model trained on CDR-H3 sequences to analyze antibody repertoires, capturing immunological signatures and supporting classifier development for broadly neutralizing antibodies.

The paper fine-tunes a protein language model using synthetic metalloprotein sequences to improve recovery of metal-binding motifs and identifies novel metal-binding sites.

The paper shows that Apaf1 forms transient foci during apoptosis, which correspond to the apoptosome and regulate apoptosis progression.

This paper develops a deep learning framework to generate light chains from heavy chains using unpaired antibody sequences, revealing maturation-linked pairing patterns.

This paper evaluates how design choices in chemical language models affect their performance and interpretability, finding that molecular representation and tokenization strategies significantly influence internal representations.

The paper uses Molecular Dynamics simulations and AlphaFold2 to study PhoQ signaling, revealing three conformations and insights into its activation mechanism.

This study evaluates AlphaFold2's ability to predict protein structures with limited sequence information, revealing its reliance on template structures and local biophysical energy functions.

This paper introduces TemBERTure, a Deep Learning framework for predicting protein thermostability using sequence data, showing improved performance with diverse datasets and attention mechanisms.