Spaces:

---

Honzus24

/

flexpert

like 0

Running on Zero

Dear Hugging Face Team,

We are a team of researchers led by Josef Sivic at the Czech Technical University in Prague. In our recent ICLR 2025 paper, we introduced a model to predict per-residue protein flexibility. That is, given a protein sequence or sequence and structure, our model predicts how much each amino acid in the protein will be moving within the protein structure.

Predicting protein flexibility computationally is critical, as it allows us to design more stable and efficient enzymes for industrial and therapeutic applications without relying solely on expensive and time-consuming wet-lab experiments.

The academic community of structural biologists is already using our flexibility predictor. For example, researchers use our predictor to elucidate critical enzymatic regions in an ultra-high-throughput screening experiment of dehalogenase enzymes. The engineering of the critical regions of dehalogenes is a way to develop improved catalysts that find applications in the biodegradation of toxic compounds in the environment. Or, researchers use our predictor to identify promising mutants of a potential protein-based thrombolytic drug, which could become a cost-effective alternative for treating patients with acute stroke.

According to the feedback we received from our users, it would be much appreciated if our tool could be used as a user-friendly web server. To make our tool more widely available, we have therefore implemented it in Hugging Face Spaces to eliminate the need for our users to install all dependencies and to provide a user-friendly interface. Our current version of this web server is functional, yet impractical, as our model would greatly benefit from the availability of a GPU.

We would therefore like to use this opportunity to apply for the GPU community grant, which could enable using this Hugging Face Space for real-world impact.

Thank you for considering our application!
Best regards,

Jan Richtr, Petr Kouba, Josef Sivic, Jiri Sedlar