A New Normalization Method on Metagenomic Sequencing Data

• Activity Number: 393
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Ruofei Du*+ and Lingling An
• Companies: University of Arizona and University of Arizona
• Keywords: next generation sequencing ; normalization ; metagenomics

Abstract

Metagenomic data, generated by high-throughput sequencing technologies, have been dramatically expanding our knowledge of microbial communities residing in environments under different conditions. Normalization plays a critical role in comparing multiple metagenomic communities. A widely used normalization method in metagenomic study is to divide the number of read counts for each feature (e.g., gene or species) by the total number of reads in the sample. However, this straightforward technique introduces biases in metagenomic differential analysis. We demonstrate the biases are likely due to the complex abundance structure of features in the microbial community and uneven sequencing depth. Due to the differences in data characteristic and structure, assumptions suitable for microarray or RNA-seq data are not valid to metagenomic data any more. We propose a new method that combines both scaling and distribution adjustment. Compared with other available normalization methods the proposed method demonstrates better performance through comprehensive simulation studies. The new method is also applied on real metagenomic datasets.