Abstract
High-throughput sequencing technologies, in particular next-generation sequencing (NGS) technologies, have emerged as the preferred approach for exploring both gene function and pathway organization. Data from NGS technologies pose new computational and statistical challenges because of their massive size, limited replicate information, large number of genes (high-dimensionality), and discrete form. They are more complex than data from previous high-throughput technologies such as microarrays. In this work we focus on the statistical issues in analyzing and modeling NGS data for selecting genes suitable for further exploration and present a brief review of the relevant statistical methods. We discuss visualization methods to assess the suitability of statistical models for these data, statistical methods for modeling differential gene expression, and methods for checking goodness of fit of the models for NGS data. We also outline areas for further research, especially in the computational, statistical, and visualization aspects of such data.
Keywords
Clustering, differential gene expression, dimension reduction, generalized linear models, hierarchical Bayesian modeling, microarrays, next-generation sequencing, negative binomial distribution, Poisson distribution, residual plots
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Recommended Citation
Srivastava, Sanvesh and Doerge, R. W.
(2012).
"THE NUANCES OF STATISTICALLY ANALYZING NEXT-GENERATION SEQUENCING DATA,"
Conference on Applied Statistics in Agriculture.
https://doi.org/10.4148/2475-7772.1038
THE NUANCES OF STATISTICALLY ANALYZING NEXT-GENERATION SEQUENCING DATA
High-throughput sequencing technologies, in particular next-generation sequencing (NGS) technologies, have emerged as the preferred approach for exploring both gene function and pathway organization. Data from NGS technologies pose new computational and statistical challenges because of their massive size, limited replicate information, large number of genes (high-dimensionality), and discrete form. They are more complex than data from previous high-throughput technologies such as microarrays. In this work we focus on the statistical issues in analyzing and modeling NGS data for selecting genes suitable for further exploration and present a brief review of the relevant statistical methods. We discuss visualization methods to assess the suitability of statistical models for these data, statistical methods for modeling differential gene expression, and methods for checking goodness of fit of the models for NGS data. We also outline areas for further research, especially in the computational, statistical, and visualization aspects of such data.
