Genotyping-by-Sequencing SNP Identification for Crops without a Reference Genome: Using Transcriptome Based Mapping as an Alternative Strategy

Next-generation sequencing opens the way for genomic studies of diversity even for non-model crops and animals. Genome reduction techniques are becoming progressively more popular as they allow a fraction of the genome to be sequenced for multiple individuals and/or populations. These techniques are an efficient way to explore genome diversity in non-model crops and animals for which no reference genome is available. Genome reduction techniques emerged with the development of specific pipelines such as UNEAK (Universal Network Enabled Analysis Kit) and Stacks. However, even for non-model crops and animals, transcriptomes are easier to obtain, thereby making it possible to directly map reads. We investigate the direct use of transcriptome as an alternative strategy. Our specific objective was to compare SNPs obtained from the UNEAK pipeline as well as SNPs obtained by directly mapping genotyping-by-sequencing reads on a transcriptome. We assessed the feasibility of both SNP datasets, UNEAK and transcriptome mapping, to investigate the diversity of 91 samples of wild pearl millet sampled across its distribution area. Both approaches produced several tens of thousands of single nucleotide variants, but differed in the way the variants were identified, leading to differences in the frequency spectrum associated with marked differences in the assessment of diversity. Difference in the frequency spectrum significantly biased a large set of diversity analyses as well as detection of selection approaches. However, whatever the approach, we found very similar inference of genetic structure, with three major genetic groups from West, Central, and East Africa. For non-model crops, using transcriptome data as a reference is thus a particularly promising way to obtain a more thorough analysis of datasets generated using genome reduction techniques.