NABench: Large-Scale Benchmarks of Nucleotide Foundation Models for Fitness Prediction

Zhongmin Li, Runze Ma, Jiahao Tan, TAN Chengzi, Shuangjia Zheng

Keywords: Foundation model, Fitness prediction, Nucleic acid

TL;DR: We present a large-scale evaluation of nucleotide foundation models across diverse tasks, providing deeper insights into nucleic acid fitness prediction.

Abstract: Nucleotide sequence variation can induce significant shifts in functional fitness. Recent nucleotide foundation models promise to predict such fitness effects directly from sequence, yet heterogeneous datasets and inconsistent preprocessing make it difficult to compare methods fairly across DNA and RNA families. We introduce NABench, a large-scale, systematic benchmark for nucleic acid fitness prediction. NABench aggregates 162 high-throughput assays and curates 2.6 million mutated sequences spanning diverse DNA and RNA families, with standardized splits and rich metadata. We show that NABench surpasses prior nucleotide fitness benchmarks in scale, diversity, and data quality. Under a unified evaluation suite, we rigorously assess 29 representative sequence models across zero-shot, few-shot prediction, transfer learning, and supervised settings. The results quantify performance heterogeneity across tasks and nucleic-acid types, demonstrating clear strengths and failure modes for different modeling choices and establishing strong, reproducible baselines. We release NABench to catalyze progress in nucleic-acid modeling and to support downstream applications in nucleotide molecular design, synthetic biology, and biochemistry. Our code is available at https://anonymous.4open.science/r/NABench-20CB.

Primary Area: datasets and benchmarks

Submission Number: 9519