Strategies for the intelligent integration of genetic variance information in multiscale models of neurodegenerative diseases

Abstract

A more complete understanding of the genetic architecture of complex traits and diseases can maximize the utility of human genetics in disease screening, diagnosis, prognosis, and therapy. Undoubtedly, the identification of genetic variants linked to polygenic and complex diseases is of supreme interest for clinicians, geneticists, patients, and the public. Furthermore, determining how genetic variants affect an individual’s health and transmuting this knowledge into the development of new medicine can revolutionize the treatment of most common deleterious diseases. However, this requires the correlation of genetic variants with specific diseases, and accurate functional assessment of genetic variation in human DNA sequencing studies is still a nontrivial challenge in clinical genomics. Assigning functional consequences and clinical significances to genetic variants is an important step in human genome interpretation. The translation of the genetic variants into functional molecular mechanisms is essential in disease pathogenesis and, eventually in therapy design. <br /> Although various statistical methods are helpful to short-list the genetic variants for fine-mapping investigation, demonstrating their role in molecular mechanism requires knowledge of functional consequences. This undoubtedly requires comprehensive investigation. Experimental interpretation of all the observed genetic variants is still impractical. Thus, the prediction of functional and regulatory consequences of the genetic variants using in-silico approaches is an important step in the discovery of clinically actionable knowledge. Since the interactions between phenotypes and genotypes are multi-layered and biologically complex. Such associations present several challenges and simultaneously offer many opportunities to design new protocols for in-silico variant evaluation strategies. <br /> This thesis presents a comprehensive protocol based on a causal reasoning algorithm that harvests and integrates multifaceted genetic and biomedical knowledge with various types of entities from several resources and repositories to understand how genetic variants perturb molecular interaction, and initiate a disease mechanism.<br /> Firstly, as a case study of genetic susceptibility loci of Alzheimer’s disease, I reviewed and summarized all the existing methodologies for Genome Wide Association Studies (GWAS) interpretation, currently available algorithms, and computable modelling approaches. In addition, I formulated a new approach for modelling and simulations of genetic regulatory networks as an extension of the syntax of the Biological Expression Language (OpenBEL). This could allow the representation of genetic variation information in cause-and-effect models to predict the functional consequences of disease-associated genetic variants. Secondly, by using the new syntax of OpenBEL, I generated an OpenBEL model for Alzheimer´s Disease (AD) together with genetic variants including their DNA, RNA or protein position, variant type and associated allele. To better understand the role of genetic variants in a disease context, I subsequently tried to predict the consequences of genetic variation based on the functional context provided by the network model. I further explained that how genetic variation information could help to identify candidate molecular mechanisms for aetiologically complex diseases such as Alzheimer’s disease (AD) and Parkinson’s disease (PD). Though integration of genetic variation information can enhance the evidence base for shared pathophysiology pathways in complex diseases, I have addressed to one of the key questions, namely the role of shared genetic variants to initiate shared molecular mechanisms between neurodegenerative diseases. I systematically analysed shared genetic variation information of AD and PD and mapped them to find shared molecular aetiology between neurodegenerative diseases. <br /> My methodology highlighted that a comprehensive understanding of genetic variation needs integration and analysis of all omics data, in order to build a joint model to capture all datasets concurrently. Moreover genomic loci should be considered to investigate the effects of GWAS variants rather than an individual genetic variant, which is hard to predict in a biologically complex molecular mechanism, predominantly to investigate shared pathology.