Become a colleague in my department!
Applications close in four days, December 15.

Alignment of each motif concept peaks at the position of that motif in the protein. We also detect a few motifs absent from individual databases, though these are typically annotated in other databases. 4/4

Building on the idea of Concept Activation Vectors from arxiv.org/pdf/1711.11279 3/4

We take embeddings of protein fragments w/ and w/o a motif, train a simple linear classifier, and use the normal vector to the decision boundary as the “motif direction.” So for motif detection, all you need is a dictionary of learned motif concept vectors, and a PLM to embed the protein with. 2/4

Alignment profiles of several motifs scanning across a protein. Each line represents a concept activation vector from a different layer.

Vision models have directions in embedding space for concepts like “stripes” or "corgi"
We show that protein language has directions for motifs, use this as a new way to detect and localize motifs!
New preprint w/@ahmadshamail.bsky.social
Feedback very welcome arxiv.org/abs/2511.21614 1/4