一种用于加速基因组分析的增强型Borzoi
Johannes C Hingerl1,2, Alexander Karollus1,2, Julien Gagneur1,2,3,4
1School of Computation, Information and Technology, Technical University of Munich, Munich, Germany.
Bioinformatics (Oxford, England)
|September 4, 2025
概括
Flashzoi 增强了 Borzoi 基因组学模型,使用旋转定位编码和 FlashAttention-2 进行更快的计算. 这加快了DNA序列,基因调控和遗传变异效应的分析.
科学领域:
- 基因组学
- 计算生物学
- 生物信息学
背景情况:
- 预测DNA序列在基因调节和变异效应中的作用在基因组学中至关重要.
- 博尔佐模型是调控基因组学的基础,分析了许多基因组测试.
- 目前Borzoi的相对位置编码阻碍了计算效率.
研究的目的:
- 提高Borzoi模型的计算效率.
- 保持或提高基因组测定和变异效应的准确性.
- 促进大规模的基因组分析和探索复杂的监管机制.
主要方法:
- 开发了Flashzoi,一个增强的Borzoi模型.
- 实现了旋转位置编码.
- 集成的FlashAttention-2可以提高计算性能.
主要成果:
- 实现了三倍以上的训练和推断速度.
- 减少了多达 2.4 倍的内存使用量.
- 在RNA-seq覆盖模型,变异效应预测和增强剂-促进剂链接中保持或提高准确性.
结论:
- Flashzoi显著提高了Borzoi在基因组分析中的效率.
- 模型的速度和减少的内存足迹使得更大规模的研究成为可能.
- Flashzoi支持对基因调节和复杂基因组建模的进一步研究.
相关概念视频
FISH - Fluorescent In-situ Hybridization
21.1K
Fluorescence in situ hybridization, or FISH, was developed in the early 1980s and has quickly become one of the most widely used techniques in cytogenetics. Labeled probes are used to bind complementary DNA or RNA sequences on a chromosome or in a region within a cell. Earlier, the probes could only be obtained by cloning or reverse transcription of a DNA template. Currently, the probe oligonucleotides can be synthesized synthetically. Additionally, with the advancement of optical techniques,...
21.1K
Genomics
37.4K
Genomics is the science of genomes: it is the study of all the genetic material of an organism. In humans, the genome consists of information carried in 23 pairs of chromosomes in the nucleus, as well as mitochondrial DNA. In genomics, both coding and non-coding DNA is sequenced and analyzed. Genomics allows a better understanding of all living things, their evolution, and their diversity. It has a myriad of uses: for example, to build phylogenetic trees, to improve productivity and...
37.4K
Animal Mitochondrial Genetics
8.0K
Among all the organelles in an animal cell, only mitochondria have their own independent genomes. Animal mitochondrial DNA is a double-stranded, closed-circular molecule with around 20,000 base pairs. Mitochondrial DNA is unique in that one of its two strands, the heavy, or H, -strand is guanine rich, whereas the complementary strand is cytosine rich and called the light, or L, -strand. Compared to nuclear DNA, mitochondrial DNA has a very low percentage of non-coding regions and is marked by...
8.0K


