Résumé
Lignocellulosic biomass from cultivated poplars is a renewable resource of interest for the production of bio-energy. To improve this complex trait, it is essential to better understand the genetic factors that affect its variation. Within this context, to explore the components of genetic variability within a parental species of commercial hybrid poplars, we have initiated an integrative approach combining genomic, transcriptomic and phenotypic information in a large collection of individuals sampled from natural populations and raised in a clonal field trial.Here, we report the analysis of polymorphism within transcriptomic sequences produced by RNA-seq. RNA was extracted from young differentiating xylem and cambium collected on 24 3-years old trees corresponding to 2 replicates of 12 genotypes from 6 natural populations representative of the geographical distribution of the species in Western Europe.After RNA quantification, the RNA from xylem and cambium from the same tree were pooled and subjected to high-throughput sequencing on an Illumina HiSeq 2000 using 100 bp paired-end indexing runs. All the samples were sequenced on one Illumina flow-cell, yielding around 18 million of pairedend reads per sample.The sequences were first trimmed (quality and length) and aligned to the Populus trichocarpa v3 assembly using BWA-MEM. Around 99.7% of the reads mapped against the reference genome and 93.3% properly mapped despite the fact that the genome is from a different species. Then, we performed a mapping filtering for duplicated sequence reads after alignment to the genome with Picard tools to mitigate biases introduced by data generation steps such as PCR amplification. We also performed local realignment around indels and recalibrated the base quality scores with GATK. The variant discovery process was based on several variant callers (GATK, VarScan, FreeBayes) with several filtering parameters (VCF tools). We compared the genotype calls to those previously obtained with a SNP array (Faivre-Rampant et al, 2016) in order to choose the optimal variant caller and the SNP filtering parameters (depth, quality).The usefulness and relevance of the resulting SNPs for population genomic studies was further assessed by analyzing: (i) genomewide patterns of nucleotide diversity, (ii) population structure, and (iii) linkage disequilibrium. The SNPs have also been annotated in order to estimate Ka/Ks ratio, with a particular focus on candidate genes involved in the control of lignocellulosic biomass production and quality.This promising approach iscurrently being extended to 240 genotypes in order to validate the genomics patterns identified in the present study.