Meeting Abstract

33.4  Friday, Jan. 4  Nonparametric bootstrapping of RNASeq data in polymorphic polyps of Hydractinia symbiolongicarpus SANDERS, S.M.*; SHCHEGLOVITOVA, M; CARTWRIGHT, P; University of Kansas; University of Kansas; University of Kansas s743s088@ku.edu

One of the most common utilities of RNASeq data is comparative gene expression between different tissues or individuals. These differences in expression data can be correlated to differences in environmental effects, morphology, and developmental stages. A colony of Hydractinia symbiolongicarpus comprises genetically identical yet morphologically distinct and functionally specialized polyp types. In order to identify differential expression patterns in these different polyp types, RNASeq libraries were constructed from feeding (gastrozooids), reproductive (gonozooids), and defensive (dactylozooids) polyps for Illumina sequencing. A hybrid transcriptome was assembled on the pooled Illumina reads. The raw reads of the individual libraries were then mapped back to the hybrid transcriptome to obtain expression counts of transcripts from different polyp types. Current differential expression packages measure statistical significance of differential expression by obtaining point estimates from expression data for known distributions (Poisson, Negative Binomial, etc.). With few or no technical replicates (extraction, library construction or sequencing) these methods have little power to discover significant results. In the absence of technical replicates, we used a nonparametric bootstrapping algorithm to assess confidence in potential differentially expressed transcripts. The bootstrapping algorithm simulates data under a binomial sampling distribution where the probability of sampling an individual transcript is equal to the number of times that transcript was found in the library. Confidence intervals on expression counts were taken for each transcript in each library to discern significant differential abundance. Here we report differentially expressed genes in different polyps types and the utility of applying nonparametric bootstrap methods to RNASeq data.