Peptide markers of aminoacyl tRNA synthetases facilitate taxa counting in metagenomic data
1 School of Physics and Astronomy, Tel Aviv University, Tel Aviv 69978, Israel
2 Sackler Faculty of Medicine and the Blavatnik School of Computer Sciences, Tel Aviv University, Tel Aviv 69978, Israel
BMC Genomics 2012, 13:65 doi:10.1186/1471-2164-13-65Published: 10 February 2012
Taxa counting is a major problem faced by analysis of metagenomic data. The most popular method relies on analysis of 16S rRNA sequences, but some studies employ also protein based analyses. It would be advantageous to have a method that is applicable directly to short sequences, of the kind extracted from samples in modern metagenomic research. This is achieved by the technique proposed here.
We employ specific peptides, deduced from aminoacyl tRNA synthetases, as markers for the occurrence of single genes in data. Sequences carrying these markers are aligned and compared with each other to provide a lower limit for taxa counts in metagenomic data. The method is compared with 16S rRNA searches on a set of known genomes. The taxa counting problem is analyzed mathematically and a heuristic algorithm is proposed. When applied to genomic contigs of a recent human gut microbiome study, the taxa counting method provides information on numbers of different species and strains. We then apply our method to short read data and demonstrate how it can be calibrated to cope with errors. Comparison to known databases leads to estimates of the percentage of novelties, and the type of phyla involved.
A major advantage of our method is its simplicity: it relies on searching sequences for the occurrence of just 4000 specific peptides belonging to the S61 subgroup of aaRS enzymes. When compared to other methods, it provides additional insight into the taxonomic contents of metagenomic data. Furthermore, it can be directly applied to short read data, avoiding the need for genomic contig reconstruction, and taking into account short reads that are otherwise discarded as singletons. Hence it is very suitable for a fast analysis of next generation sequencing data.