With the advances in medical genomics, it has been shown that many genetic disorders previously considered to be monogenic, may be attributed to more complex inheritance mechanisms, following instead an oligogenic inheritance model. However, little is still known about the genetic causes of these disorders. The aim of this work is the study of digenic diseases, the simplest case of oligogenic disorders, and the construction of predictive methods that can distinguish variant combinations within two genes leading to disease or not. For this purpose, we exploited the information present in the publicly available DIDA database, whose main entity is a digenic combination (i.e. a combination of variants within two genes) leading to a digenic disorder, combined with information of the involved genes and their associated genetic variants. As a neutral dataset, we used the variant information of healthy individuals from the 1000 genome project, further filtered and annotated to create comparable digenic combinations with those in DIDA. Using these instances, a random forest predictor for digenic combinations was created. Our results reveal that single variant effect predictors on the gene and protein function (such as Polyphen-2) together with Pfam information, as well as differences in the wild type and mutated amino acid properties, are essential for the discrimination of neutral from disease-causing digenic combinations. These results constitute a first step in determining the genetic causes of digenic diseases and open the path for the construction of more advanced predictive tools for complex genetic disorders.
Papadimitriou, S, Gazzo, A, Smits, G, Nowé, A & Lenaerts, T 2016, 'Predicting digenic variant effects with DIDA', 15th European Conference on Computational Biology, The Hague, Netherlands, 3/09/16 - 7/09/16.
Papadimitriou, S., Gazzo, A., Smits, G., Nowé, A., & Lenaerts, T. (2016). Predicting digenic variant effects with DIDA. Poster session presented at 15th European Conference on Computational Biology, The Hague, Netherlands.
@conference{006fdfcfd3c34d75809aca5772ea23f1,
title = "Predicting digenic variant effects with DIDA",
abstract = "With the advances in medical genomics, it has been shown that many genetic disorders previously considered to be monogenic, may be attributed to more complex inheritance mechanisms, following instead an oligogenic inheritance model. However, little is still known about the genetic causes of these disorders. The aim of this work is the study of digenic diseases, the simplest case of oligogenic disorders, and the construction of predictive methods that can distinguish variant combinations within two genes leading to disease or not. For this purpose, we exploited the information present in the publicly available DIDA database, whose main entity is a digenic combination (i.e. a combination of variants within two genes) leading to a digenic disorder, combined with information of the involved genes and their associated genetic variants. As a neutral dataset, we used the variant information of healthy individuals from the 1000 genome project, further filtered and annotated to create comparable digenic combinations with those in DIDA. Using these instances, a random forest predictor for digenic combinations was created. Our results reveal that single variant effect predictors on the gene and protein function (such as Polyphen-2) together with Pfam information, as well as differences in the wild type and mutated amino acid properties, are essential for the discrimination of neutral from disease-causing digenic combinations. These results constitute a first step in determining the genetic causes of digenic diseases and open the path for the construction of more advanced predictive tools for complex genetic disorders.",
author = "Sofia Papadimitriou and Andrea Gazzo and Guillaume Smits and Ann Now{\'e} and Tom Lenaerts",
year = "2016",
month = sep,
day = "3",
language = "English",
note = "15th European Conference on Computational Biology, ECCB2016 ; Conference date: 03-09-2016 Through 07-09-2016",
}