We want to genotype a panel using SeqSNP (a targeted GBS technology). We have designed over 2k probes targeting some SNPs positions across the genome, but we can only select 1k probes for the genotyping. We tried to sort these probes based on chromosomes and positions (bp) (ascending order), determined the distances between probes and set a threshold to eliminate probes that are less than the threshold. However, we think that we could miss some regions whilst having a cluster in some other regions.
Is there any other criteria we could use to ensure we evenly distribute the probes across the genome?
Having been involved in designing SNP chips for diploid species over many years, these are some factors we took into account. First was the likelihood that the SNP actually existed and was not a sequencing error, the presence of both alleles in a number of individuals is a good filter. Second was that depending on technology used there was no variants or repeat regions in methods that required oligo based recognition sites. Third was equal spacing, although the assumption of 1Mbp=1cM is probably flawed and so a bit higher density near telomeres is advised. The fourth is the minor allele frequency estimate, but this is nuanced. At low densities you probably have to select for MAF >0.3. At higher densities you should probably treat each region of the genome as a segment and select SNPs with a range of allele frequencies and/or minimise the LD between them. Personally though I would only use SeqSNP (We use the GT-seq variant in-house) for targeted variants or where you want to maintain compatibility with say an existing parentage plex. For many applications: genomic selection, parentage, linkage and LD mapping and genomic diversity studies methods like restriction enzyme reduced representational sequencing aka GBS or ddRADseq are much more flexible and cheaper unless you are doing tens of thousands of samples see https://www.researchgate.net/project/Genotyping-for-production-and-security-in-a-biological-economy-an-MBIE-project Hope this helps.
- Badges
- Science topic
- Similar topics
- Correction