Mona Shojaei, Navid Mohammadvand, Tunca Doğan, Can Alkan, Rengül Çetin Atalay, Aybar C. Acar
We developed a sequence-based method, Pathogenic Mutation Prediction (PMPred), for predicting the pathological nature of variants in the human proteome based on analysis of gene ontologies. Indeed, PMPred calculates the pathogenicity rating of mutations that significantly disturb protein function or lead to loss of protein function. Since nonsense and frameshift mutations are more likely than missense mutations and in-frame indels to alter or shorten protein sequences, they may more significantly truncate protein function. This method measures the significance of variants effect on the function and represents the outcomes in decreasing order of pathogenicity. As the scale of pathogenicity, we provided Pearson correlation, ranging between 0 and ± 1. Then, if the comparison gives a value near 1, there is a perfect correlation even after alteration. In order to assess the harmful effect of mutations value of 0.9 is recommended as the cutoff between benign and pathogenic variants. Additionally, domain prediction results and information regarding the location of mutations are displayed to assist the user with better data interpretation.
The documentation is available at Github.
If you fail to retrieve the prediction results, your input file might have too many variants. Since gene ontology-based functional analysis takes a while, it is advised to use small sets that contains most suspicious variants thought to be related to the target disease. Also, if the uploaded file is not appropriately formatted as a VCF or has an unsupported file extension, you will not be able to get the results.
Please note that we are currently supporting GRCh37/hg19. If you are retrieving scores for only a subset of variants the reason could be that you're sending in variants that are based on a different genomic build.
monashojaei@metu.edu.tr
Informatics Institute
Middle East Technical University Ankara, 06800 Turkey