COMAN: a web server for comprehensive metatranscriptomics analysis

Abstract

Background: Microbiota-oriented studies based on metagenomic or metatranscriptomic sequencing have revolutionised our understanding on microbial ecology and the roles of both clinical and environmental microbes. The analysis of massive metatranscriptomic data requires extensive computational resources, a collection of bioinformatics tools and expertise in programming.

Results: We developed COMAN (Comprehensive Metatranscriptomics Analysis), a web-based tool dedicated to automatically and comprehensively analysing metatranscriptomic data. COMAN pipeline includes quality control of raw reads, removal of reads derived from non-coding RNA, followed by functional annotation, comparative statistical analysis, pathway enrichment analysis, co-expression network analysis and high-quality visualisation. The essential data generated by COMAN are also provided in tabular format for additional analysis and integration with other software. The web server has an easy-to-use interface and detailed instructions, and is freely available at http://sbb.hku.hk/COMAN/ CONCLUSIONS: COMAN is an integrated web server dedicated to comprehensive functional analysis of metatranscriptomic data, translating massive amount of reads to data tables and high-standard figures. It is expected to facilitate the researchers with less expertise in bioinformatics in answering microbiota-related biological questions and to increase the accessibility and interpretation of microbiota RNA-Seq data.

Keywords: Computational biology; Metatranscriptomics; Microbial RNA-Seq; Microbial community; Web servers.

Fig. 1

The metatranscriptome analysis pipeline in COMAN. The example output figures from the analysis of a gut microbiome dataset are shown. Upper left: functional profiling; upper right: taxonomic contribution analysis (complete linkage method for clustering algorithm and Euclidean distance for dissimilarity metric); lower left: multi-dimensional scaling to illustrate sample clustering; lower right: co-expression network analysis with different inferred communities (for clarity purpose, the communities with fewer than 3 elements are merged together and not highlighted here)

Fig. 2

Performance evaluation of subsets of the combined database used in COMAN. The combined database was constructed by merging the NCBI bacterial reference genomes non-coding RNAs with eukaryotic ribosomal DNA (both large and small subunits) deposited in the SILVA database. Different subsets of random 10 % and 5 % of the full combined database (indicated by x-axis) were taken and their performance was compared to the BLASTN mapping results from using the full version. For each subset, the “Relative Accuracy” is defined as the number of commonly identified reads between the subset and the full database, divided by the total number of reads identified only by using the subset for mapping. In comparison, the “Relative Sensitivity” is defined as the number of commonly identified reads between the subset and the full database, divided by the total number of reads identified by using the full database