Whole Genome Shotgun Sequencing Metagenomics Analysis for the "Common" Scientist

Introduction: Whole Genome Shotgun Sequencing (WGS) of the microbiota is a powerful technique that provides an unbiased look at the relative abundance of the gut microbiome, but the analysis of the large data sets requires a bioinformatician using software executed on high-powered servers. This led to development of metagenomics analysis software aimed at reducing computing resources. These programs can allow the biological scientist to perform metagenomics analysis using standard personal computers (PC).

Objective: Compare performance of selected metagenomics analysis software on a chicken cecal microbiota WGS data set with a PC.

Methods: All software were executed on a Lenovo ThinkCentre M93 with Intel Core i7 4770 with 32GB RAM running Fedora (Heisenberg). Raw 150-bp pair-end reads from Illumina HiSeq 2500 were pre-processed with FastX. The final data set used was approximately 291 million reads totaling just over 100GB spread across 8 FASTQ files. Four metagenomics analysis software were executed: Taxy, GRAMMy, MetaPhlAn, and MEGAN. Additionally, PAUDA and DIAMOND were used to generate the BlastX inputs required by MEGAN.

Results: Taxy failed to execute due to MATLAB coding that was not supported by Octave. GRAMMy’s execution requires listing of reference genomes, which introduces bias to the relative abundance estimations. Executing PAUDA and DIAMOND with NCBI’s nr database was too much for the PC, resulting in out-of-memory error (PAUDA) and excessive runtime (DIAMOND). Attempt using NCBI’s env_nr database produced nonsensical results from MEGAN. MetaPhlAn was the only software that successfully predicted relative abundances on the PC.