SNP calling vs. individual curated genes using 454 data from multiple heterozygotes
Hi.. yes, the title says it all. I need to identify novel SNPs (or, for that matter, any polymorphisms) in a bunch of curated genes (1500 or so). We don't have a genome. I have a lot of 454 exome sequencing data, which is specifically enriched for the genes I am looking at; this comes from a number of individuals in a mapping population.
So far, I have de novo assembled the 454 data (to ensure that the gene models we have are valid, but that's a different problem). I've also had a go at mapping the raw 454 reads (from one individual) using an individual gene as a backbone. Obvious SNPs there. This has been done using both MIRA and Geneious. MIRA produces a file detailing where it thinks there are SNPs, but I don't want to re-invent the wheel by writing a program to convert it into a .gff file if it's been done already, and there's no statistical "niceness" score for the SNPs.
So a few questions..
1)Is there any software that can call SNPs/polymorphisms and give me some statistical measure of the "goodness" of the SNP, i.e. the likelihood that the SNP is correct and not just a sequencing artefact?
2) Is there any software that can call SNPs and produce an annotation track ( .gff file) that I can then annotate the gene models with?
3) Is the above a good approach and am I missing anything?
Any tips on automating this process would be appreciated too. I've already written scripts to automate the de novo assembly / mapping to individual genes, but tips on how to select good SNPs from the mapping etc would be nice.
Thanks all from New Zealand!
So far, I have de novo assembled the 454 data (to ensure that the gene models we have are valid, but that's a different problem). I've also had a go at mapping the raw 454 reads (from one individual) using an individual gene as a backbone. Obvious SNPs there. This has been done using both MIRA and Geneious. MIRA produces a file detailing where it thinks there are SNPs, but I don't want to re-invent the wheel by writing a program to convert it into a .gff file if it's been done already, and there's no statistical "niceness" score for the SNPs.
So a few questions..
1)Is there any software that can call SNPs/polymorphisms and give me some statistical measure of the "goodness" of the SNP, i.e. the likelihood that the SNP is correct and not just a sequencing artefact?
2) Is there any software that can call SNPs and produce an annotation track ( .gff file) that I can then annotate the gene models with?
3) Is the above a good approach and am I missing anything?
Any tips on automating this process would be appreciated too. I've already written scripts to automate the de novo assembly / mapping to individual genes, but tips on how to select good SNPs from the mapping etc would be nice.
Thanks all from New Zealand!
Comment