Standard Post Variant Call (Vcf) Analysis That Work Out Of The Box

Question

[William]

Lets say that you have whole genome variant calls for a number of individuals from a population within a species.

What are standard analysis / tools that you can always run out of the box after you get a a multi-sample vcf file?

• SNPEff -> Effect prediction, Ts/Tv ratio, Codon changes, Amino acid changes, Changes by chromosome
• SNPSifft - > filtering the SNPEff annotated data
• VCF-tools VCF-stats -> Number of private SNP per individual
• SNPRelate -> PCA plot, Phylogentic tree

I guess a lot of people are nowadays sequencing multiple individuals from a population and they all want to explore the variant calls / relation / contrast between individuals. Of course experiment, questions and populations are different but it is always good to minimize reinventing the wheel and maximizing usage of existing tools.

Things I am for example looking for are:

• plotting variant densities
• plotting LD plots

Answer 1

[Emily]

What about the Ensembl Variant Effect Predictor (VEP). It allows you to input a VCF file and get a list of the genes/transcripts affected, the amino acid change (where relevant) along with SIFT and Polyphen scores and protein domains hit, tells you if the variants are already known and gives you their ID and can give you publications where the variant was seen, allows you to show or even filter by population frequency (in species where this data is available) as well as a bunch of other options. It's available as a standalone perl script which can be used offline if you set up a species cache, and as a simple web tool, although the web tool is only for ≤750 variants so wouldn't be suitable for whole genome calls.

[William]

To be honest, I don't really trust anything written in perl, things become a big unreadable unmaintainable mess easily. If there is a Java (snpEff) or C++ alternative, I'll go with that one.

[Bert Overduin]

So, you select your tools by the language they're written in? Frankly, this is the most ridiculous thing I've heard in a long time. When it comes to writing your own code I fully understand that people have their preferences, but judging a tool by the language it's written in instead of how well it works, really. Especially if you say you're interested in tools that work out of the box, the language in which the tool was written should be of no concern to you. How good a tool is doesn't depend on the language it has been written in, but how good a job the people did that wrote the code. Maybe not every developer is capable of writing good Perl code, but our developers certainly are.

[Emily]

You know what they say about bad workmen...

[Alex Paciorkowski]

Oh good grief, and I don't trust anything written in whitespace, that becomes unreadable even quicker :P The web-based VEP is a great tool, use it every day.

Answer 2

[Jorge Amigo]

using ANNOVAR locally you'll be able to annotate your variants using virtually any UCSC's track, plus some other useful resources as described on the web. for instance, mining for allele frequencies on 1000genomes, ESP or COSMIC has always helped us in our experiments.

Answer 3

[alexej.knaus]

Use GeneTalk to analyze human variants and talk about it.

• register at www.gene-talk.de for free
• upload VCF files onto your account
• your VCF files will be annotated and during the preprocessing step a quality assesment will be performed (if file contains more than 10.000 lines)
• filter your file with the filter tools
• take a look at annotations, comment or rate them according to medical relevance and scientific eveidence, create annotations on questionable variants
• take a look at the gene where that has a questinable variant, see if experts are available that might help you in interpreting the variant.
• share your expertise with the community or ask experts

in the following white paper you will find more information about the platform and some of its tools and features: Documentation Area at GeneTalk

• First Steps with GeneTalk
• Working with Annotations
• Compound heterozygous filtering

[Adrian Pelin]

Is this possible on non models?

[alexej.knaus]

sorry, but I do not understand your question,

GeneTalk is only available vor human variant files...