TL;DR
This paper introduces PLATCOM, a flexible system for comparative genomics that integrates various computational tools and databases. It addresses the challenge of analyzing massive genomic datasets by providing a user-friendly environment for exploring relationships between multiple genomes, with a focus on sequence analysis and data mining.
What Problem Does This Paper Address?
The exponential growth of genomic sequence data presents a significant challenge for researchers. Analyzing and comparing multiple genomes is crucial for understanding evolutionary processes and identifying common biological features. However, this task is computationally demanding and requires integrating a multitude of specialized tools and diverse data sources. Existing systems often lack flexibility, require extensive programming knowledge, or present results in a way that is difficult to interpret. PLATCOM aims to bridge this gap by offering a unified, flexible, and user-friendly platform for comparative genomics, specifically from a "sequence analysis perspective."
Methods
PLATCOM is built around a modular architecture that integrates internal databases (GenBank, COG, KEGG) and a custom Pairwise Comparison Database (PCDB). PCDB stores pre-computed all-against-all comparisons of protein and whole genome sequences, significantly speeding up analysis. The system incorporates a suite of widely used public sequence analysis tools (e.g., FASTA, BLASTZ, HMMER) alongside custom-developed high-performance data mining tools. These tools are organized into genome analysis modules, designed to be combinable and reconfigurable. Key modules include:
- Genome Plot: Generates 2D diagonal plots for visualizing pairwise genome comparisons and identifying conserved gene clusters.
- Operon Analysis (OperonViz): Navigates and visualizes gene neighborhoods, aiding in the identification of horizontal gene transfers and functional coupling.
- Gene Fusion Event Detection (FuzFinder): Identifies plausible gene fusion events by analyzing mutual best hits between genes in different genomes.
- Metabolic Pathway Analysis (MetaPath): Reconstructs metabolic pathways by combining KEGG information with sequence data, aiming to identify missing genes.
- Gene Clustering Tools (e.g., FASTA-BAG, Genome-BAG): Utilizes algorithms like BAG for clustering sequences and identifying relationships.
- Genome Alignment Tools (GAME): Provides pairwise and multiple genome alignment capabilities.
A core design principle is the development of abstract data types for genome comparison and a formal approach to defining operations and function compositions, paving the way for a genome analysis language.
Datasets
The paper mentions the use of several public databases as internal data sources for PLATCOM:
- GenBank: (ftp://ftp.ncbi.nlm.nih.gov/genomes)
- SwissProt: (http://www.ebi.ac.uk/swissprot)
- COG (Clusters of Orthologous Groups): (http://www.ncbi.nlm.nih.gov/COG)
- KEGG (Kyoto Encyclopedia of Genes and Genomes): (http://www.genome.ad.jp/kegg)
The Pairwise Comparison Database (PCDB) is a custom database containing 97,034 entries of pairwise comparisons of protein sequence files (.faa) and whole genome sequence files (.fna) from 312 replicons.
Key Insights and Ideas
- Pre-computation is Key for Speed: Storing pre-computed pairwise comparisons in PCDB dramatically accelerates genome analysis, making it feasible for web-based applications.
- Modularity and Reconfigurability: Designing modules with well-defined interfaces allows for flexible combination and reuse, enabling users to construct custom multi-step analyses.
- Data Mining as a Unifying Force: High-performance data mining tools are essential for integrating disparate sequence analysis tools and databases, effectively "gluing" them together at the biological sequence level.
- Abstract Data Types for Formalization: Defining abstract data types for genomes, sequences, and their relationships provides a foundation for a more formal and extensible genome analysis language.
- Visualization for Interpretation: Interactive visualization tools are crucial for summarizing complex analysis results and providing users with intuitive insights into genomic relationships.
Key Results
The paper highlights that PLATCOM runs "several hundred times faster than a system without PCDB" for multiple genome comparisons. The system is designed to be flexible, extensible, scalable, and reconfigurable, with an emphasis on high-performance data mining. The authors describe the completion of the first stage of PLATCOM's development, focusing on the underlying architecture and individual system modules.
Limitations and Future Work
- Limited Data Types: While the paper proposes abstract data types, the current implementation might not encompass all possible data types for genome comparison.
- Prokaryotic Focus for MetaPath: The MetaPath web service is currently limited to prokaryotic genomes.
- Directionality in Metabolic Pathways: The directionality of metabolic pathways is not yet considered in MetaPath due to limitations in KEGG data.
- Expertise for Module Combination: While PLATCOM offers modularity, combining modules for complex multi-step analyses still requires substantial programming practice, limiting its accessibility to bioinformatics experts. The authors aim to address this by developing an environment where biologists can perform multi-step analyses flexibly.
- Genome Analysis Language: The genome analysis language for PLATCOM is described as being in its "infant stage," with future work focused on formalizing function definitions and enriching the syntax and semantics.
Summary updated