Objective

Of the twenty-nine species in the genus Populus, black poplar (P. nigra L.) is native to Eurasia with a wide distribution area including Europe, as well as the southwest and central Asia, and northwest Africa [1]. It is regarded as a keystone species for riparian ecosystems in ecological and conservation studies [2] and it has an interest as a parental pool in interspecific (P. deltoides x P. nigra) poplar breeding programs as the origin of cultivated hybrids [3].

In this study, we selected 241 genotypes from a P. nigra collection of 587 genotypes previously genotyped with an Illumina 12 K Infinium Bead-Chip array (8000 Single Nucleotide Variants (SNVs); [4]), which in turn belong to a larger collection of 1098 cloned genotypes sampled in natural populations from 11 river catchments [5] in four European countries. These 241 genotypes were previously studied for wood properties in 2 sites (Savigliano-2011 & Orléans-2012; [6]), and their selection was based on the following set of criteria defined following a first analysis of population structure with 8000 SNVs: (i) introgression < 10% of the worldwide-spread fastigiated form P. nigra var italica, (ii) proportion of recruitment to their ancestral population > 50% and (iii) survival in the common garden located at Orléans, France.

We generated transcriptomics data from young differentiating xylem and cambium tissues of these P. nigra selected genotypes by using RNAseq technology. We launched on the resulting sequences an in-silico pipeline [7] that allowed us to obtain 878,957 polymorphisms. We already used this data in Chateigner et al. [8] and in Wade et al. [9] for phenotype prediction. These data provide new valuable resources for a wide variety of genome-based studies, ranging from population structure analysis over distribution ranges, to genomic prediction and genome-wide association studies (GWAS) for traits related to wood properties and growth, for example.

Data description

Young differentiating xylem and cambium tissues were harvested in June 2015 from 480 Populus nigra trees from a common garden located at Orléans (France) (241 genotypes, 2 clonal trees per genotype, Data file 1 [10], Data set 1 [11]). RNA from the xylem and cambium were extracted with RNeasy Plant kit (Qiagen, France) according to manufacturer’s recommendations. Treatment with DNase I (Qiagen, France) was carried out to ensure elimination of genomic DNA. RNA was eluted in RNAse-DNAse free water and quantified with a Nanodrop spectrophotometer. RNA from xylem and cambium of the same plant were pooled in an equimolar extract (250 ng/μL) and sent to the sequencing platform. The sequencing platform POPS (transcriptOmic Platform of Institute of Plant Sciences—Paris-Saclay) prepared the RNAseq libraries from polyA-RNA selection using the TruSeq_Stranded_mRNA_SamplePrep_Guide_15031047_D protocol (Illumina, California, U.S.A.).

Identity of each sample was checked with the following procedure: a first round of variant detection was performed with FreeBayes (v.1.0.0) [7] with small modifications to detect SNVs. All experiment steps (from growth conditions to bioinformatic analyses) are available in the CATdb database (Data set 3 [15]). Briefly, the reads were first trimmed with Trimmomatic (v.0.38) [16] to remove adapter and low-quality sequences. Then, they were aligned to the Populus trichocarpa reference genome v.3.0 [17] using the BWA-MEM algorithm (v.0.7.12) [19, 20] for RNAseq short variant discovery: we first marked the duplicates with the MarkDuplicates from the Picard tools (v.2.0.1) [21] and then used the SplitNCigarReads, the Indel Realignment and the Base Quality Recalibration tools from GATK (v.3.5) [22]. SNV and short insertions and deletions were genotyped for all the sequenced trees (the same genotypes were pooled together) with 3 variants callers: (i) GATK using the HaplotypeCaller tool in single-sample calling mode followed by joint genoty** of the samples with the GenotypeGVCFs tool; (ii) FreeBayes (v.1.0.0) [24].

The resulting 3 files (one per caller) were filtered with VCFtools (v.0.1.15) [25] to obtain only the intra-specific (Populus nigra) biallelic SNV with a variant quality score (QUAL) threshold over 30. These filtered files were then combined together with the vcf-isec tool from VCFtools and we only kept the genotype calls that were detected by at least 2 variant callers. Otherwise, the genotype call was set as a missing value for this particular individual. As there remained some missing data (Data file 3 [26]), genotype imputation was performed using the Fimpute program (v.2.2) [27] for the SNV located on the chromosomes/scaffolds that contain at least 2 SNV. Thereby, 9.73% of missing values were imputed. This yielded genotypes at 878,957 biallelic sites for the 241 Populus nigra individuals (Data file 4 [28]) without missing data. 878,893 of them have an annotation (Data file 5 [29], Data file 6 [30]) from ANNOVAR (v.2017Jul16) [31] (Data file 7 [32]). Among them, 868,861 are located on the 19 chromosomes (Data file 7, [32]). A total of 26,909 genes harbored between 1 and 309 SNVs.

Limitations

  • The SNVs are limited to expressed genes: 26,909 vs a maximum of 41,335 P. trichocarpa protein-coding genes. We found a correlation between expression level and SNV density at gene level, which is likely due to an increase in coverage for highly expressed genes (Data file 7 [32]).

  • RNA were extracted from young differentiating xylem and cambium tissues, which might not be representative of transcriptional activity of other tissues of the tree. The found SNVs are therefore specific to the genes expressed in these two tissues. Nevertheless, this makes this SNV dataset clearly appropriate to study the genetics of wood formation.

  • The SNVs are limited to the gene space: 92.55% of the found SNV are exonic, intronic, in the 3’UTR (UnTranslated Region) or in the 5’UTR (Data file 7 [32]).