[论文解读] Deep Recurrent Neural Network for Protein Function Prediction from Sequence

Xueliang Liu|arXiv (Cornell University)|Jan 28, 2017

Machine Learning in Bioinformatics参考文献 16被引用 13

一句话总结

该论文提出一种带有长短期记忆（LSTM）单元的深度循环神经网络（RNN），直接从氨基酸序列预测蛋白质功能，无需序列比对或特征工程。在UniProt数据上进行训练后，该模型在同类别和跨类别预测中均表现出高准确率，包括对已实验验证的、与已知同源物序列相似性较低的铁氧还蛋白样蛋白的发现。

ABSTRACT

As high-throughput biological sequencing becomes faster and cheaper, the need to extract useful information from sequencing becomes ever more paramount, often limited by low-throughput experimental characterizations. For proteins, accurate prediction of their functions directly from their primary amino-acid sequences has been a long standing challenge. Here, machine learning using artificial recurrent neural networks (RNN) was applied towards classification of protein function directly from primary sequence without sequence alignment, heuristic scoring or feature engineering. The RNN models containing long-short-term-memory (LSTM) units trained on public, annotated datasets from UniProt achieved high performance for in-class prediction of four important protein functions tested, particularly compared to other machine learning algorithms using sequence-derived protein features. RNN models were used also for out-of-class predictions of phylogenetically distinct protein families with similar functions, including proteins of the CRISPR-associated nuclease, ferritin-like iron storage and cytochrome P450 families. Applying the trained RNN models on the partially unannotated UniRef100 database predicted not only candidates validated by existing annotations but also currently unannotated sequences. Some RNN predictions for the ferritin-like iron sequestering function were experimentally validated, even though their sequences differ significantly from known, characterized proteins and from each other and cannot be easily predicted using popular bioinformatics methods. As sequencing and experimental characterization data increases rapidly, the machine-learning approach based on RNN could be useful for discovery and prediction of homologues for a wide range of protein functions.

研究动机与目标

• 解决从蛋白质一级氨基酸序列预测功能的挑战，而无需依赖序列比对或人工特征工程。
• 开发一种机器学习方法，使其在具有相似功能但系统发育关系较远的蛋白质家族间具有泛化能力。
• 实现对未标注蛋白的发现，这些蛋白可能与已知功能类别具有潜在的功能相似性。
• 在实验上未表征的序列上验证模型的预测能力，特别是那些与已知功能同源物序列相似性较低的序列。

提出的方法

• 使用带有长短期记忆（LSTM）单元的深度循环神经网络（RNN）架构，直接从蛋白质一级序列处理序列型氨基酸数据。
• 模型在UniProt的公开标注蛋白数据集上进行端到端训练，学习与功能类别相关的序列模式的分层表征。
• 未使用序列比对或启发式评分；相反，原始氨基酸序列被嵌入并作为序列标记进行处理。
• RNN学习序列中的长程依赖关系，使得即使在局部不保守的情况下也能识别功能基序。
• 模型在四种关键蛋白质功能上进行微调和评估：CRISPR相关核酸酶、铁氧还蛋白样铁储存蛋白、细胞色素P450及其他功能。
• 通过利用训练数据中学到的功能相似性，对系统发育上相距较远的家族进行跨类别预测。

实验结果

研究问题

• RQ1 带有LSTM单元的深度RNN能否在不进行序列比对或特征工程的情况下，直接从一级氨基酸序列准确预测蛋白质功能？
• RQ2 该模型在系统发育关系较远但功能相似的跨类别蛋白质家族上的泛化能力如何？
• RQ3 该模型能否识别并预测大型数据库（如UniRef100）中未标注序列的功能？
• RQ4 该模型在识别与已知功能蛋白序列相似性较低的新功能同源物方面，其发现能力在多大程度上有效？
• RQ5 RNN预测的功能是否可经实验验证，特别是在传统方法失效的情况下？

主要发现

• RNN模型在四种主要蛋白质功能的同类别预测中表现优异，优于依赖手工设计序列特征的其他机器学习方法。
• 该模型成功预测了与已知同源物序列相似性较低的蛋白质中具有铁氧还蛋白样铁隔离功能，并经实验验证。
• 跨类别预测在系统发育上相距较远的家族中识别出功能相似的蛋白质，包括CRISPR相关核酸酶和细胞色素P450，尽管序列相似性较低。
• 在部分未标注的UniRef100数据库上应用时，该模型预测了已知和新发现的候选蛋白，包括此前未标注目标功能的序列。
• 该模型发现了铁氧还蛋白样家族中未被标准BLAST或其他传统生物信息学工具检测到的功能同源物。