Development of an improved Polygenic Risk Scores (PRS) for complex disorders
Approved Research ID: 93016
Approval date: September 21st 2022
Advances in genomic medicine and analysis tools for the prediction of some diseases through genetic biomarkers have led to improvements in the analysis of the polygenic risk score (PRS). This is achieved by combining the effect size of several SNPs. In many diseases these PRS have already been identified (i.e. neurogenerative diseases, heart disease, breast cancer, diabetes, etc.). In diseases such as schizophrenia, the PRS is able to explain approximately 8% of the risk of developing schizophrenia. Moreover, we know that this early prediction can improve patient care through prevention, active follow-up and genetic counseling consultations, as demonstrated by the study in Finnish population and risk of cardiovascular disease with atherosclerosis. Therefore, prediction using PRS has been shown to increase the ability to predict risk to certain diseases over the classical single SNP approach. The same has been seen in the case of breast cancer, whose prediction by PRS, allows to classify patients into percentiles according to risk, and this has allowed in some countries its incorporation in the clinical routine for breast cancer screening.
However, the sensitivity and specificity of some algorithms, as well as the lack of including some additional variables, make necessary the development of other algorithms that can improve the existing ones. These algorithms are based on different mathematical models. It should also be taken into account that some PRS can be specific to some populations, so the previous analysis of ethnicities, using different tools, can improve these prediction algorithms.
In addition, the computation of these algorithms may require high computational capacities and CPU memory usage, which makes it necessary to develop more efficient algorithms that are able to perform these calculations and predictions in increasingly shorter times.
Therefore, the aim of this project is to develop, test and validate, using the UK biobank cohort, an algorithm to estimate the genetic risk of several diseases that do not yet have it, as well as to improve some existing algorithms.
ADNTRO has a database of more than 10,000 samples from consenting individuals to develop an algorithm for predicting susceptibility to various diseases. Therefore, the main objective is to validate this algorithm with the UKbiobank data, publish the results and make this algorithm freely available to all researchers and professionals who may find it useful.