Development of a universal and simplified ddRAD library preparation approach for SNP discovery and genotyping in angiosperm plants

Abstract

BackgroundThe double digest restriction-site associated DNA sequencing technology (ddRAD-seq) is a reduced representation sequencing technology by sampling genome-wide enzyme loci developed on the basis of next-generation sequencing. ddRAD-seq has been widely applied to SNP marker development and genotyping on animals, especially on marine animals as the original ddRAD protocol is mainly built and trained based on animal data. However, wide application of ddRAD-seq technology in plant species has not been achieved so far. Here, we aim to develop an optimized ddRAD library preparation protocol be accessible to most angiosperm plant species without much startup pre-experiment and costs.ResultsWe first tested several combinations of enzymes by in silico analysis of 23 plant species covering 17 families of angiosperm and 1 family of bryophyta and found AvaII + MspI enzyme pair produced consistently higher number of fragments in a broad range of plant species. Then we removed two purifying and one quantifying steps of the original protocol, replaced expensive consumables and apparatuses by conventional experimental apparatuses. Besides, we shortened P1 adapter from 37 to 25 bp and designed a new barcode-adapter system containing 20 pairs of barcodes of varying length. This is an optimized ddRAD strategy for angiosperm plants that is economical, time-saving and requires little technical expertise or investment in laboratory equipment. We refer to this simplified protocol as MiddRAD and we demonstrated the utility and flexibility of our approach by resolving phylogenetic relationships of two genera of woody bamboos (Dendrocalamus and Phyllostachys). Overall our results provide empirical evidence for using this method on different model and non-model plants to produce consistent data.ConclusionsAs MiddRAD adopts an enzyme pair that works for a broad range of angiosperm plants, simplifies library constructing procedure and requires less DNA input, it will greatly facilitate designing a ddRAD project. Our optimization of this method may make ddRAD be widely used in fields of plant population genetics, phylogenetics, phylogeography and molecular breeding.Electronic supplementary materialThe online version of this article (doi:10.1186/s13007-016-0139-1) contains supplementary material, which is available to authorized users.

Optimizing ddRADseq in Non-Model Species: A Case Study in Eucalyptus dunnii Maiden

“…The addition of 24 adapters with barcodes in the ligation step allowed the pooling of samples and the application of size selection before PCR, as reported in ddRADseq [9]. Moreover, the use of different barcode lengths (4 to 9 bp, Poland et al [27]) allowed us to avoid sequencing phasing error at the beginning of reads, as reported in GBS and MiddRADseq [3,17], but not considered in the original ddRADseq [9]. For this scaled-up P2, we also proposed the use of automatic size selection, which would decrease the possibility of cross-contamination and increase the precision and consistency when applying the protocol for more than one pool of samples [5], as reported in the original ddRADseq [9].…”

“…Ligation: 24 variable-length (4 to 9 bp) barcodes designed by Poland et al [30] were added, in order to avoid low sequence quality of the first bases due to the restriction site [3,17] (Supplementary Table S2).…”

“…The original ddRADseq protocol was built and trained based on animal data, and it has been widely applied in SNP marker development and genotyping for several species in this kingdom. Applications of this technology to plants have been reported [10][11][12][13][14], especially in forest and fruit trees (reviewed in [15]), and were specifically improved [16,17]. However, most researchers still use universal default protocols that carry some limitations.…”

Double-digest RAD-sequencing: do wet and dry protocol parameters impact biological results?

“…2) was required in our library preparation for the two tested systems, which could open the possibility of using ddRADseq technique with low quantity DNA template from non-invasive sampling in a conservation genomics context. If the initial protocol from Peterson et al (2012) already suggested to use less than 100 ng of DNA per individual, ddRADseq users commonly process more than 200 ng and even up to 1 μg (Capblancq et al, 2015; Yang et al, 2016; Burns et al, 2017; Sherpa, Rioux, Goindin, et al, 2018). Here we showed that enzyme saturation can already happen at the digestion step with 250 ng.…”

“…This highlights the trade-off existing between a satisfactory coverage, directly related to the number of PCR cycles, and the limitation of errors occurring during PCR, because these can lead to very weak fragment coverage impeding loci reconstruction (Hohenlohe et al, 2012). Usually the number of PCR cycles is set between 12 and 16 (Peterson et al, 2012; Capblancq et al, 2015; Yang et al, 2016; Burns et al, 2017). Considering the important increase in individual heterozygosity and number of private alleles observed with 25 PCR cycles, our results greatly support this practice.…”