Augmenting Biological Fitness Prediction Benchmarks with Landscapes Features from GraphFLA
Mingyu Huang, Shasha Zhou, Ke Li · 2025
Read Augmenting Biological Fitness Prediction Benchmarks with Landscapes Features from GraphFLA by Mingyu Huang, Shasha Zhou, Ke Li (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-24826.txt.
Introduction
Mingyu Huang, Shasha Zhou, and Ke Li add shape facts to bench sets that test how well models guess the fitness of a gene or protein form. Such bench sets draw on lab data but lack a map of how rough or smooth the trait land is. With no such map, a mean score hides why a model wins or fails.
Their tool, GraphFLA, builds trait lands from change data for DNA, RNA, and protein, up to vast counts of forms. It counts twenty traits that mark four sides of land shape. Run on more than five thousand lands, it showed which traits sway model scores. That helps teams read past a single mean and pick models with eyes open.
Catalog record
| CREATOR | Mingyu Huang, Shasha Zhou, Ke Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 155 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-24826