arXiv cs.AI· Zhang Liang, Zhang Yukun, Tan Yang, Jiang Fan, Zhong Bozitao, Cheng Xiaoran, Hong Liang, Li Mingchen, Yu Yuanxi·· 11 小时前AI 评分27
VenusGenome：面向微生物基因组表示学习的基因组感知掩码自编码器
VenusGenome: Genome-Aware Masked Autoencoding for Microbial Genome Representation Learning
AI 导读
VenusGenome 是一个基因分辨率掩码自编码器，结合仅可见编码与融入 contig 边界、相对链方向和基因位置的基因组感知注意力，在超 170 万个微生物基因组组装上预训练。其冻结表示在 11 项主要指标中的 9 项上取得所评估模型的最高平均性能，消融实验支持基因组感知注意力的贡献，并显示掩码自编码优于 BERT 式掩码建模、更高掩码比例带来进一步提升。
正文 · 原文
Abstract:Genomic-context models have shown that contextualizing protein embeddings can improve predictions of microbial function. We observe that a gene's protein embedding is partially predictable from the embeddings of neighboring genes in the genome, motivating high-ratio masking that requires reconstruction from sparser genomic context. Genomic modeling must also distinguish biological relationships from assembly conventions, since the ordering of separate contigs does not imply adjacency and strand labels depend on the reference orientation. We therefore introduce VenusGenome, a gene-resolution masked autoencoder that combines visible-only encoding with genome-aware attention incorporating contig boundaries, relative strand orientations, and gene positions. We pretrain this model on a curated corpus of over 1.7 million microbial genome assemblies from isolate and metagenomic sources, using frozen protein embeddings as gene-level inputs. Across genome-level phenotype prediction and gene-level tasks, the resulting frozen representations achieve the highest mean performance among the evaluated models on nine of eleven primary metrics. An exploratory gene-input perturbation analysis further links representation sensitivity to experimentally measured fitness defects. Ablations support the contribution of genome-aware attention and show the benefits of masked autoencoding over BERT-style masked modeling, with further gains from higher masking ratios.
| Subjects: | Genomics (q-bio.GN); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2611.05405 [q-bio.GN] |
| (or arXiv:2611.05405v1 [q-bio.GN] for this version) | |
| https://doi.org/10.48550/arXiv.2611.05405 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Liang Zhang [view email]
[v1]
Tue, 20 Oct 2026 09:40:01 UTC (12,468 KB)
来源：arXiv cs.AI · arxiv.org