[论文解读] PhyloPythiaS+: A self-training method for the rapid reconstruction of low-ranking taxonomic bins from metagenomes

Ivan Gregor, Johannes Dröge+3 more|arXiv (Cornell University)|Jun 27, 2014

Genomics and Phylogenetic Studies参考文献 51被引用 17

一句话总结

PhyloPythiaS+ 引入了一种自训练方法，可自动化构建基于组成特征的宏基因组分类分类器，以自动管道替代人工专家校准。该方法使 k-mer 计数速度提升 100 倍，总运行时间减少三倍，并可在低成本硬件上实现完全自动化、高精度地从 Gb 级宏基因组数据中重建物种和属级别的分组。

ABSTRACT

Metagenomics is an approach for characterizing environmental microbial communities in situ, it allows their functional and taxonomic characterization and to recover sequences from uncultured taxa. For communities of up to medium diversity, e.g. excluding environments such as soil, this is often achieved by a combination of sequence assembly and binning, where sequences are grouped into 'bins' representing taxa of the underlying microbial community from which they originate. Assignment to low-ranking taxonomic bins is an important challenge for binning methods as is scalability to Gb-sized datasets generated with deep sequencing techniques. One of the best available methods for the recovery of species bins from an individual metagenome sample is the expert-trained PhyloPythiaS package, where a human expert decides on the taxa to incorporate in a composition-based taxonomic metagenome classifier and identifies the 'training' sequences using marker genes directly from the sample. Due to the manual effort involved, this approach does not scale to multiple metagenome samples and requires substantial expertise, which researchers who are new to the area may not have. With these challenges in mind, we have developed PhyloPythiaS+, a successor to our previously described method PhyloPythia(S). The newly developed + component performs the work previously done by the human expert. PhyloPythiaS+ also includes a new k-mer counting algorithm, which accelerated k-mer counting 100-fold and reduced the overall execution time of the software by a factor of three. Our software allows to analyze Gb-sized metagenomes with inexpensive hardware, and to recover species or genera-level bins with low error rates in a fully automated fashion.

研究动机与目标

• 消除宏基因组分类分组中对人工专家校准的需求。
• 实现大规模（Gb 级）宏基因组数据集的可扩展、自动化分析。
• 在无需专业技能的情况下实现高精度的物种和属级别分组。
• 开发一种自训练框架，能够从宏基因组数据本身学习，降低对外部参考数据库的依赖。
• 显著减少微生物群落研究中分类分组的计算时间和资源需求。

提出的方法

• 该方法采用自训练管道，能自动从宏基因组样本中识别标记基因以定义分类分组。
• 通过数据驱动方法识别具有分类学信息的序列，替代人工专家选择训练序列的角色。
• 一种新型 k-mer 计数算法使 k-mer 频率计算速度提升 100 倍，大幅缩短整体运行时间。
• 该软件采用基于组成特征的分类方法，利用 k-mer 频率将序列分配至分类分组。
• 该系统设计为可在标准硬件上高效运行，使其在微生物组研究中可广泛使用。
• 支持从原始测序数据到分类分组的全流程自动化，最大限度减少用户干预。

实验结果

研究问题

• RQ1 自训练方法能否替代人工专家校准，用于构建宏基因组的分类分类器？
• RQ2 在不损害分组准确性的前提下，k-mer 计数最多可加速多少？
• RQ3 自动化分组能否在大型宏基因组数据集中实现物种和属级别的分辨率且错误率较低？
• RQ4 仅使用标准计算硬件是否可行实现高精度、低排名的分类分组？
• RQ5 与专家校准方法（如 PhyloPythiaS）相比，自训练分类器在准确性和速度方面表现如何？

主要发现

• PhyloPythiaS+ 中的自训练方法成功替代了人工专家校准，实现了分组流程的完全自动化。
• 新型 k-mer 计数算法使 k-mer 频率计算速度提升 100 倍。
• 与原始的 PhyloPythiaS 相比，软件的整体执行时间减少了三倍。
• 该方法可在廉价硬件上分析 Gb 级宏基因组，显著提升了可扩展性。
• 低排名分类分组（物种和属级别）以低错误率被成功恢复，证明了自动化分组的高准确性。
• 该软件在保持与专家校准方法相当的高性能和准确性的同时，消除了对专业技能的需求。