Study : Characterizing restriction enzyme-associated loci in historic ragweed (Ambrosia artemisiifolia) voucher specimens using custom-designed RNA probes
Characterizing restriction enzyme-associated loci in historic ragweed (Ambrosia artemisiifolia) voucher specimens using custom-designed RNA probes
Population genetic studies of non-model organisms frequently employ reduced representation library (RRL) methodologies, many of which rely on protocols in which genomic DNA is digested by one or more restriction enzymes. However, because high molecular weight DNA is a prerequisite for these protocols, samples with degraded DNA are generally unsuitable for RRL methods. Given that ancient and historic specimens can provide key temporal perspectives to evolutionary questions, we explored how custom-designed RNA probes could enrich for RRL loci (Restriction Enzyme-Associated Loci baits, or REALbaits). Starting with Genotyping-by-Sequencing (GBS) data generated on modern common ragweed (Ambrosia artemisiifolia L.) specimens, we designed 20,000 RNA probes to target well-characterized genomic loci in herbarium voucher specimens dating from 1835–1913. Compared to shotgun sequencing, we observed enrichment of the targeted loci at 20–117-fold. Using our GBS capture pipeline on a dataset of 38 herbarium samples, we discovered 20,944 SNPs, providing sufficient genomic resolution to distinguish geographic populations. For these samples, we found that dilution of REALbaits to 10% of their original concentration still yielded sufficient data for downstream analyses and that a sequencing depth of ~7.5M reads was sufficient to characterize most loci without wasting sequencing capacity. In addition, we observed targeted loci had highly variable rates of success, which we primarily attribute to similarity between loci, a trait that ultimately interferes with unambiguous read mapping. Our findings can help researchers design capture experiments for RRL loci, thereby providing an efficient means to integrate samples with degraded DNA into existing RRL datasets.