I am planning to genotype ~520 samples using original RADseq (Etter et al. 2011). The project has 2 goals: (1) construct a linkage map using a subset of those individuals, and (2) examine introgression among populations. I would like feedback on my sequencing plan.
I am studying two hybridizing species with a genome size of ~980 Mb. We do not have whole genome sequence and suspect low sequence divergence between the species. We would like to generate as many tags as possible so we can make a dense linkage map.
An in silico digest of transcriptome data from a closely related species suggests that the restriction enyme BbvCI will yield ~120k RAD tags while allowing multiplexing of 85 samples per lane (at 30x coverage on an Illumina HiSeq), for a total of 6 lanes. For comparison, a less frequent cutter like SfbI would allow us to sequence everything on 2 lanes, but would yield only ¼ the number of tags. BbvCI was the only enzyme I tested that is estimated to yield a large number of fragments while also allowing an acceptable level of multiplexing, which is why I would like to use it. We have sufficient funding for more sequencing lanes, but one feature of BbvCI makes me question its usefulness:
BbvCI does not have a palindromic cut site, so each DNA fragment will have a different 5’ overhang at each end: either TCA or TGA. Does this mean that I need to order 2 different P1 adapters, 1 for each end? If I were to use only one P1 adapters and ligate it to one end, this would halve the number of RAD tags generated.
I welcome any suggestions on how to optimize the tradeoff between number of tags generated and costs of oligos, enzymes, and sequencing.
I am studying two hybridizing species with a genome size of ~980 Mb. We do not have whole genome sequence and suspect low sequence divergence between the species. We would like to generate as many tags as possible so we can make a dense linkage map.
An in silico digest of transcriptome data from a closely related species suggests that the restriction enyme BbvCI will yield ~120k RAD tags while allowing multiplexing of 85 samples per lane (at 30x coverage on an Illumina HiSeq), for a total of 6 lanes. For comparison, a less frequent cutter like SfbI would allow us to sequence everything on 2 lanes, but would yield only ¼ the number of tags. BbvCI was the only enzyme I tested that is estimated to yield a large number of fragments while also allowing an acceptable level of multiplexing, which is why I would like to use it. We have sufficient funding for more sequencing lanes, but one feature of BbvCI makes me question its usefulness:
BbvCI does not have a palindromic cut site, so each DNA fragment will have a different 5’ overhang at each end: either TCA or TGA. Does this mean that I need to order 2 different P1 adapters, 1 for each end? If I were to use only one P1 adapters and ligate it to one end, this would halve the number of RAD tags generated.
I welcome any suggestions on how to optimize the tradeoff between number of tags generated and costs of oligos, enzymes, and sequencing.
Comment