Discrete Kernel Density Estimation for RNA-Sequence Data Sets

• Activity Number: 320
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Samuel Benidt*
• Companies: Iowa State University
• Keywords: Gene Expression ; RNA-seq

Abstract

With the glut of statistical methods for RNA-seq data available in the literature, researchers are interested in simulation studies that compare the performance of these methods. The efficacy of these methods for RNA-seq data is often studied through simulation studies based on the negative binomial distribution. While the negative binomial distribution is quite flexible in modeling count data, it may not accurately model the gene dependence structure of RNA-seq data, leading to a misleading view of the performance of a given method when applied to real RNA-seq data.

To more accurately model the complex structure of RNA-seq data, we propose a kernel-density based estimator of the read count distribution of a given gene using discrete associated kernels. Further applications of the kernel density estimator are discussed.