Back to catalogue >>

Reference: Tian J., Wu N., Guo X., Guo J. Zhang J., Fan Y. Predicting the phenotypic effects of non-synonymous single nucleotide polymorphisms based on support vector machines. BMC Bioinformatics (2007) 8 450-464.
Hosted: Hosted at the Molecular Biology and Bioinformatics department of the Biotechnology Research Institute, Chinese Academy of Agricultural Sciences, Beijing. (http://www.mobioinfor.cn/parepro/contact.htm)

Only available for download and running locally on the command line.

Summary:
Parepro identifies variants with a deleterious effect on protein function using a support vector machine (SVM).

Methodology:
Parepro is trained on the HumVar mutation data implemented in the PhD-SNP and Hansa algorithms. This dataset comprises 13032 disease-related substitutions from 1111 genes and 8946 neutral substitutions from 3484 genes. Three attributes are characterised from homologues collected by a PSI-BLAST:
• The residue differences - the property differences between the ‘new’ amino acid and those in the alignment.
• The mutation position information - the distribution of amino acids at the position.
• The sequence environment - the amino acids upstream and downstream from the mutation site.

Input:
• The user must provide a multiple sequence alignment featuring the protein of interest and homologues.
• The following executables must be run:
 psap.exe –i alignment_file –o psap_output_file –t protein_ID
 pare_pre.exe –i psap_output_file –p substitution_position –w wildtype_aa –r variant_aa –v vector_file
• This will then display the details of the substitution, whether it’s a neutral variant or not, and a reliability score.