Herbert Wertheim College of Engineering
Fan Lab
Led by Xiao Fan · Biomedical Engineering
What we study
The Fan Lab uses computer models and statistical methods to study rare diseases. They analyze genetic data to find disease causes and develop treatments based on those findings.
What working here is like
- Computational
- No organisms (data only)
- Coding required
- Projects
- 2
- Papers
- 5
What we're looking for
“Students with good programming skills are encouraged to contact me directly.”
Current projects
SHINE
Uses protein language models to predict pathogenicity of short inframe insertion and deletion variants.
Large Language Modeling Predictions of Missense Variant Pathogenicity
The lab builds AI models to guess whether genetic changes cause disease. Using 680,079 variants (about 7% pathogenic) for training and testing on 10,784 variants from four genes, the student will train a neural network with protein‑language embeddings and compare results from different models.
From the Spring 2025 UF Undergraduate Research Symposium · abstract book, p. 542
Summaries are written by AI in plain language from public symposium abstracts and may be imperfect.
Recent papers, explained
- Structural basis for the assembly and modulation of human M-channels
2026 · Vita
Maps how human M-channels form and how drugs affect them.
- Conformational switch of NPC1 epitomizes a lumen-to-membrane alternating access model for nonpolar substrates
2026 · Proceedings of the National Academy of Sciences
Shows how a protein moves cholesterol in a disease-causing protein complex.
Show 3 more papersShow fewer
- Loss of ptr-6 restores eggshell integrity and embryonic viability in Caenorhabditis elegans fatty acid synthase mutants
2026 · G3 Genes Genomes Genetics
Finds that removing ptr-6 protein helps eggshell and embryo survival in worms with fatty acid issues.
- Characteristics of predicated pathogenic missense variants in human genes for Mendelian disorders
2026 · Frontiers in Bioinformatics
Analyzes genetic and protein traits of harmful missense mutations in disease genes.
- PRIMED: predicting DNA binding residues by leveraging pre-trained protein language models
2026 · Frontiers in Artificial Intelligence
Builds a tool to predict DNA-binding protein spots using AI and protein data.
Plain-language summaries are AI-written and reviewed by us; check the paper for details.
How to reach out
Xiao Fan hasn't set up this page yet. You can also look them up in the UF directory (search “Xiao Fan”).