Human Gene Mutation Database

http://www.hgmd.cf.ac.uk/ac/index.php

Paid subscription for up-to-date information. Otherwise less up-to-date public version of the database is freely available only to registered users from academic institutions/non-profit organisations.

First of all, I would be hesitant using the words "causal" and "genetic variants/SNP" together. Most clinical geneticists would probably use the term "mutation" for a sequence change in a gene that clearly causes a disease. SNPs are polymorphisms putatively found in the normal population, so one would not strictly speaking talk of SNPs "causing" disease. Data linking certain combinations of SNPs with complex traits (ie GWAS data on autism, diabetes, migraine, etc) are suggestive -- but would not be spoken of as causal. The evidence is just not there for causation.

That said, there are many databases of disease-causing mutations in single genes, for example: MECP2 (Rett syndrome), SCN1A (Dravet syndrome) and many, many more...

And of course there is the HGMD that @rbagnall mentions above, that allows access to many genes at once. But often the gene-specific databases are more up-to-date as they are usually maintained by researchers actively studying that gene. Many clinical labs also have local mutation data that is not yet entered into HGMD.

Is anyone else aware of a more comprehensive genome-wide human mutation database with easily accessible phenotype information?

You could have a look at dbGaP, GWAS Central, SNPedia or ClinVar (though it is still in development). But I agree with Alex: all these databases will give you SNPs "associated" with a disease, not "causal" for it. Determining causality requires a lot more evidence than a GWA study.

Thank you for your answers.

To follow up this question:

1) I could not get HGMD to send me the password :(! Maybe I will wait another day or two before I ask them again.

2) But I found an alternative, which in part builds on OMIM. The dbSNP begins to annotate SNP with a new field called "clinical significance". This can be searched by a term such as "pathogenic[Clinical Significance]".