Missense Mutations Predictions Using Sift And Polyphen-2

Question

[Kamila001]

Hi everyone,

Is there anyone who knows if the SIFT or Polyphen-2 can be used to analyse the SNPs from pig genome?? Please I want your suggestions. I have made some changes in the uniprot files in Polyphen-2 to get the pig annotations and files but I am not sure if it gives me correct results. is there any other way to do this?

How to decide which SNPs are damaging if we get a bit different results from SIFT and Polyphen? Should I consider the overlapping set between SIFT and polyphen?

[Pierre Lindenbaum]

You should ask this directly to the authors of those softwares.

[Kamila001]

Thanks all,

I have installed Polyphen2 version 2.0.22.I think i have to upgrade this. If polyphen is specie independent then does that means I don't have to make any changes in the uniprot script to download annotations for pig?? Secondly I also run few SNPs from chicken which were predicted as deleterious by polyphen but tolerated by SIFT and vice versa e.g.

Q90808 G4R (tolerated in SIFT and deleterious in polyphen) Q3V5M1 G558R (deleterious in SIFT and benign in polyphen) Q2AB82 L295F (tolerated in SIFT and deleterious in polyphen)

So, what could be the reason when a SNP which is highly conserved across species is predicted as deleterious by SIFT but not by polyphen? How to decide which SNPs are damaging if we get a bit different results? Should I consider the overlapping set between SIFT and polyphen?

[User 59]

If this is a separate question it should be posted as such, not appended inline as an answer, or if it is a continuation of the same question please edit the original question.

I supervised a project last year on SNP phenotype prediction, and you should not expect 100% concordance between different methods. No computational method is going to tell you 100% what the effects will be, whether you choose to trust an overlapping set or not!

[Kamila001]

thanks alot Daniel for the comment.infact it was the continuation of the same question.

[Andrea_Bio]

there is a paper called 'pathogenic or not? And if so then How?' and i think i recall reading in it why you can/can't compare the results of the these tools. It certainly called for greater transparency of methodology so you could make more informed decisions on conflicts/agreement between software.

[Andrea_Bio]

the tools use different methodologies so you won't necessarily get overlap of results, and as daniel says, overlap of results doesn't confirm anything (though it might make you more willing to get the lab biologists to look into it). SIFT uses phylogenetic information and polyphen uses structural information + multiple sequence alignments. YOu really need to understand how the methods work to interpret them correctly so i think you have no option but to read their respective papers.

[Andrea_Bio]

the tools use different methodologies so the results of the 2 aren't necessarily comparable. You really need to understand the methodologies to interpret the results correctly and to decide whether you think the overlap is important. SIFT is based on phylogenetic information and polyphen is based on structural information and multiple sequence alignments. A residue might be highly conserved so sift would say a change in that is a bad thing but polyphen might look in the structure and say that the new amino acid doesn't compromise the structure.

[sruthimouli88]

Hello Kamila,

I am new to polyPhen and I am studying dog genome. I wanted to ask if u had to make any changes to the Uniprot files that you have mentioned in your question as there was no comment regarding that.

Answer 1

[Pauline Ng]

This is Pauline Ng, creator of SIFT (new website at http://sift-dna.org) Please note my possible bias in the answer to your question. :)

If you want to optimize for sensitivity, i.e. detecting all amino acids that may affect an individual, take the union of PolyPhen and SIFT.

If you want to optimize for specificity, i.e. diagnostics, take the intersection.

For your pig project, because it is a genome-wide analysis and not diagnostics, I would recommend taking the union.

Answer 2

[Andrea_Bio]

Polyphen is species independent. I've used it on cow and buffalo. Make sure to get the latest version of the software and all the latest databases otherwise you will get version conflict errors when you run it.

Answer 3

[User 59]

I think a missense mutation should be flagged as such by SIFT or Polyphen regardless of organism. There is a paper here which uses Polyphen on porcine genes if you're looking for reassurance.

Answer 4

[Lars Juhl Jensen]

I am not familiar with SIFT, but methodologically I see no reasons at all why Polyphen-2 should not work as well on pig sequences as on human sequences.

Answer 5

[Sophia]

There is also Condel , a method that calculates a combined deleteriousness score from the above mentioned methods. It can be run in a web browser from here .