RNA-Seq Mapping Software Evaluation on Non-Model Genetic Organism Gallus gallus for Differential Expression Analysis and Novel Single Nucleotide Polymorphisms (SNPs) Discovery

Introduction: With organizations such as the iPlant Collaborative (www.iplantcollaborative.org) developing user friendly interfaces like the Discovery Environment, users with minimal training can process RNA-seq data.  An early step in RNA-seq data processing is mapping, and correctly mapping the data to a reference genome can greatly influence the analysis downstream.  For non-model genetic organisms, such as chicken (Gallus gallus), performance of the mapping software is not known due to incomplete reference genome.

Objective: To characterize and evaluate popular sequence alignment software, especially those available on iPlant’s Discovery Environment.

Methods: RNA-seq data with 100-bp single-end reads from 8 individual chicken livers sequenced with Illumina HiSeq 2000 were uploaded into iPlant’s Discovery Environment, and standard RNA-seq processing procedures were carried out: pre-processing (FastX), mapping, and summarization for either differential expression analysis (Cufflinks) or novel SNPs discovery (SAMtools).  Three splice-aware mapping software (GMAP, TopHat, and STAR) and 2 non-splice-aware mapping software (Bowtie2 and BWA) were compared. Based on runtime, mapping percentage, FPKM distribution, agreement of INDELs and SNPs identified and, importantly, ease of use (evaluated by availability on iPlant, ability to run locally, multiple input sequence on iPlant, and output format).

Conclusions: For beginner users, TopHat is the ideal choice due to ease of use and has performed well in all other measurements.  For moderately experienced users, GMAP should be seriously considered as a cross-reference for the TopHat result, but runtimes are greater.  STAR was difficult to use, especially due to memory requirements for index building.