Whole genome amplification, library preparation, and sequencing
Given the small body size of CFM and the relatively high amount of input-DNA required for GBS (200 ng per sample), consistently isolating enough DNA from each specimen was challenging. Recently developed WGA methods, such as the REPLI-g WGA Mini Kit (QIAGEN) hold promise for NGS studies of small organisms. The REPLI-g mini kit uses multiple displacement amplification to amplify genomic DNA (Cheung & Nelson 1996), and typical usage can produce an average product length of 10 kb. These kits advertise uniform DNA amplification however some studies have suggested that they can introduce amplification biases, impacting genome coverage, and they have also been reported to co-amplify contaminant DNA (Ellegaard et al. 2013; de Medeiros & Farrell 2018). Although a handful of studies have used such WGA kits for NGS of small organisms (Blair et al. 2015; Onyango et al. 2015; Cruaud et al.2018; de Medeiros & Farrell 2018), only two studies have assessed the impact of amplification biases in non-pooled samples of individuals using restriction enzyme-based SNP genotyping methods, a suite of techniques that includes GBS. Blair et al. (2015) tested the effect of WGA on locus recovery and genotyping using relatively high levels of input DNA (100 ng), per manufacturer’s specifications, and reported essentially no difference in locus recovery or genotyping between treatments. A similar study using variable quantities of input DNA (as low as 6 ng) found that genome coverage appeared to be impacted by sample-specific differences in the amount of DNA used for WGA (de Medeiros & Farrell, 2018).
To test the effect of WGA on GBS approaches of small insect samples, we created GBS libraries with and without WGA for 24 of the CFM samples collected in 2017 (n = 48 libraries). Given preliminary results of these 48 libraries, the remaining 96 CFM samples collected in 2017 and 2018 underwent WGA prior to library preparation. GBS library preparation largely followed Poland et al. (2012), and any modifications to this protocol are detailed in Erlandson et al. (2019). Paired-end sequencing was conducted in two runs using an Illumina HiSeq 2500: the 24 individuals used to assess the effect of WGA on GBS library preparation were pooled and sequenced separately from the remaining 96 individuals. A 439 basepair region of the mitochondrial COI gene was also amplified for each specimen and sequenced on an ABI 3730xl Sanger sequencer following Mori et al. (2019). All sequencing (GBS and COI ) was conducted at the National Research Council of Canada Laboratory (Saskatoon, Saskatchewan, Canada).