Variation Interpretation Predictors: Principles, Types, Performance, and Choice

By Abhishek Niroula, Mauno Vihinen. 2016

120 citations recorded by the source corpus.

Next‐generation sequencing methods have revolutionized the speed of generating variation information. Sequence data have a plethora of applications and will increasingly be used for disease diagnosis. Interpretation of the identified variants is usually not possible with experimental methods. This has caused a bottleneck that many computational methods aim at addressing. Fast and efficient methods for explaining the significance and mechanisms of detected variants are required for efficient precision/personalized medicine. Computational prediction methods have been developed in three areas to address the issue. There are generic tolerance (pathogenicity) predictors for filtering harmful variants. Gene/protein/disease‐specific tools are available for some applications. Mechanism and effect‐specific computer programs aim at explaining the consequences of variations. Here, we discuss the different types of predictors and their applications. We review available variation databases and prediction methods useful for variation interpretation. We discuss how the performance of methods is assessed and summarize existing assessment studies. A brief introduction is provided to the principles of the methods developed for variation interpretation as well as guidelines for how to choose the optimal tools and where the field is heading in the future.

Subfields: Cancer Research, Genetics, Molecular Biology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cancer Research

In Genetics

In Molecular Biology

Citations in the corpus

Machine Learning in Enzyme Engineering (2019, background): “problems with binary labels or labels from a finite number of options, this evaluation is usually based on the confusion matrix: the number of true/false positives and negatives (26).”
CADD v1.7: using protein language models, regulatory CNNs and other nucleotide-level scores to improve genome-wide variant predictions (2024, background): “CADD has been widely adopted in the field of genomics and variant interpretation, aiding researchers and clinicians in prioritizing genetic variants for further analysis, especially in the context of disease-causing variants and personalized medicine applications ( 21–23 ).”
Mutation severity spectrum of rare alleles in the human genome is predictive of disease type (2019, background): “protein localization and degradation, post-translational modification, and the activity of enzymes [22, 23].”
ClinPred: Prediction Tool to Identify Disease-Relevant Nonsynonymous Single-Nucleotide Variants. (2018, unknown): “These metrics include sensitivity (true positive rate), speciﬁcity (1-false positive rate), accuracy, precision, F1 Score, and Matthew correlation coef-ﬁcient (MCC), 29,30 as well as the area under the test receiver operating characteristic curve (AUC).”