Progress Towards Standardizing Metagenomics: Applying Metagenomic Reference Material to Develop Reproducible Microbial Lysis Methods with Minimum Bias.

Metagenomics research has grown rapidly since 2010. This growth has contributed to a lack of reproducibility between different methods and laboratories, which risks limiting our ability to compare between studies and decreases confidence in previous conclusions. To address the diversity of methods available, we compared the performance of many commercially- and academically sourced lysis protocols. The methods were evaluated using mock microbial community standards with defined composition and known manufacturing tolerances to serve as a ground truth for measurement. In order to facilitate comparisons, we developed the Measurement Integrity Quotient (MIQ), providing a single, easy to understand numerical score that describes the accuracy of an observed composition relative to a known standard. Utilizing this method, we compared the effects of many different variables on lysis efficiency, including differences between thermal, enzymatic, and mechanical (bead) lysis as well as minor changes within a method, including over 40 different bead material/size combinations and cell disruptor type/intensity/run time combinations. Additionally, several replicates of typical sample types (feces, soil, skin, saliva, urine) were tested with different lysis methods with replicates at different laboratories for over 1500 samples tested. Hard, dense ceramic beads of mixed size on an appropriate cell disruptor with a validated protocol created the least biased lysis of all examined methods. The use of the MIQ score provided rapid, easy to understand analysis of accuracy and has been released in a rapidly deployable container image for public use. The use of these data, and methods can allow for the creation of more reproducible metagenomics pipelines with results that better represent the true composition of the sample. These methods were able to achieve minimal amounts of deviation from expected composition and a high degree of run-to-run and interlab reproducibility.