Hi,

I think, Bio-Informatics like data bases are one of the prime candidate in identifying a defect along with a mutation. But, as always, for any great work; you need to validate it with wet lab experiments. If the mutation falls within the functional portion of the target protein... whether there is any interacting viability that is compromised; BioInfo can give you idea. But wet labs are a must to confirm.

You may publish a Letters or a short communication using the Bio-info angle only; but to confirm an association you need to perform various in vitro/vivo experiments or observe a larger cohort to associate the mutational risk.

Hope this helps your quest.

Best.

Somnath.

It is very difficult to say for sure if some SNP variant is indeed linked to the disease in a causative way and Somnath Paul in the comments is correct that this needs to be carefully confirmed in experiments. However, SNPs that are associated (even if not in a causative way) with diseases are still of interest from a certain diagnostic perspective, as markers. The main problem you might encounter when dealing with these associations is the problem of multiple comparisons (so many putative variants can be identified that some might show up as significant hits by chance) so it is recommended to have a second association study (use an independent sample) to confirm the most promising associations. This is very trivial and you probably know all this. But here is something that might be usefull: a program that helps narrow down SNP variants that are potentially of clinical significance if they are within protein-coding regions. The program is called polyphen and it predicts "functional" mutations http://genetics.bwh.harvard.edu/pph2/. Hope this helps :)

I assume you are talking about monogenetic diseases and not association studies....

This is not a new problem, only the amount of variants you can generate has increased. And I can show you plenty of examples of failures using normal Sanger sequencing in high-ranking papers, where rare benign variants have been called pathogenic mutations.

A serious NGS department will have bioinformatic strategies for filtering variants. If you do an Exome (let alone a whole genome), you really have to. Most will automatically filter out all protein-silent exonic variants, except maybe those affecting the last base (often affecting splicing) etc.. The filter strategy will also depends on what you expect: Your "sporadic" case could mean it's a recessive disease, or it could be a de-novo mutation in a dominant disease. It helps if the disease has been defined and the mendelian pattern is known. If you expect a recessive, you only look at the genes with homozygous or compound-heterozygous variants. If you think it's dominant with de-novo mutation, a trio analysis (mother-father-index child) will be most helpful.

You can do a stepwise filter strategy, first all nonsense variants, then core-splice sites (gt/ag) and if nothing convincing shows up, look at the missense. Missense will be the most difficult of course. The bioinformatics will try to filter out all the more common variants. Databases of known or likely bening variants will get better in the  future. I like the ESP data, but study subjects in there are not fully healthy, so a variant that shows up there doesn't mean it's benign even if you look only for dominant diseases. If it shows up 20 or 30 times among the 6500 exomes and the disease you are looking at is rare, the variant most likely has nothing to do with your disease.

The process is quite well described in plenty of exome papers. But it is a lot of work, and uncertainty often remains. If you don't have experimental evidence and don't have a family to check segregation/de-novo, you can only speculate. What is known about the gene, it's domains, is there an animal model of the gene etc. If not certain, report you'r not certain!

In-silico prediction of missense variants is helpful, but by no means perfect. My favourite example is the MEFV-Gen in familial mediterranean fever, most of the well known common pathogenic mutations go as benign in Polyphen et al.....Or the other way around: Take a look at the ESP data browser, look up a gene of a known rare dominant disease. You will most likely find variants too common in the ESP to be pathogenic that PolyPhen2 will call probably-damaging (the ESP data browser shows the PolyPhen2 score for every variant....).

I   study NGS structural variations within candidate genes for genetically heterogenous complex diseases in ethnically diverse genetic isolates. Found   population-specific deletions and/or duplications, as well as common  reliable for same candidate genes in diverse isolates human  ...  looking now for same questions as following step

Thanks alot to all, for there valuable comments and suggestion,

I would like to add , if there is any bioinformatic tool which is based on wet lab experiments and can be used for determination of pathogenicity.

I agree with you Somnath Paul for the comment that need wet lab experiments to validate, But with advent of personal genome concept and use of genetics in clinical setting, we need some bioinformatics tools which not only predict but could validate the function of variation identified.

I agree with you Dr. Lars-Erik Wehner, that trio study can explain the disease mutation relationship in sporadic cases. But for those where one parent is available or no parent are there for sporadic cases and I am performing exome analysis which is kind total universe after getting variations. There should be some approach may be from system biology , that can link various mutation insilco to predict phenotype outcome.

thanks

For many monogenic forms of e.g. nonsyndromic intellectual disability animal models may not be indicative. Thus most of the papers that report about the identification of new disease genes in this context are purely based on statistics. As a rule of thumb you currently need three or more unrelated cases with a ultra rare disease with rare, potentially pathogenic mutations in the same gene. If you have a patient one of the biggest challenges is finding another one. With that taks also www.gene-talk.de might help. It's an expert network for geneticists that enables them to discuss variants of unkown clinical significance and to identify further patients of the same kind. Check it out!