I.PolyPhen SNP data collection
The developed prediction method enabled the analysis of all human non-synonymous SNPs publicly available via dbSNP database. This gave rise to a data collection of nsSNPs with predicted impact on protein structure and function.

All human SNPs from dbSNP build 126, were mapped to known human proteins. This resulted in 76,434 nsSNPs (54,573 unique) that were analysed by PolyPhen and comprise our data collection.

You can download complete dataset of PolyPhen predictions.

Alternatively, you can:

Search SNP data collection by dbSNP rs:
 

If you use the above data, please cite:

Human non-synonymous SNPs: server and survey.
Ramensky V, Bork P, Sunyaev S.
Nucleic Acids Res. (2002) 30(17): 3894-900. PubMed

See also:

dbSNP Database

PolyPhen

PolyPhen Help

  I.1.Data collection statistics

Statistics of PolyPhen predictions for dbSNP build 126

Total human refSNP entries in dbSNP...........11,961,761

54,573 unique ns rs entries in 21,405 sequences
76,434 entries in total
=======================

 1,191  unknown...............1.6%
49,795  benign...............65.2%
14,187  possibly damaging....18.6%
11,174  probably damaging....14.6%