Background / Purpose:Detecting Single Nucleotide Polymorphisms (SNPs) between genomes is a routine task with Next Generation Sequencers (NGS) data. SNP detection methods generally need a reference genome. As non-model organisms are increasingly investigated, reference-free methods are needed. The discoSnp method detects SNPs directly from raw NGS data set(s) without using any third-party information. The pea non-model organism has a 4.5 GB complex genome without reference. We compared, on the same set of low depth pea sequences, the SNPs generated by discoSnp with those published with a previous SNP discovery pipeline, and those generated using classical mapping approach with the association of Bowtie2 and GATK tools.Main conclusion:The quality of discoSnp results in association with its very low memory needs and low time footprints led us to choose this software for a SNP discovery and direct Genotypin. By Sequencing project on a set of 48 pea genomic DNA libraries from a recombinant inbred lines subpopulation sequenced with Illumina HiSeq2000 technology. The analysis enabled to identify 88,851 SNP polymorphs on this population, from which around 60k SNPs will be genetically mapped.