A machine-learning heuristic to improve gene score prediction of polygenic traits.
Ontology highlight
ABSTRACT:
SUBMITTER: Pare G
PROVIDER: S-EPMC5627249 | biostudies-literature | 2017 Oct
REPOSITORIES: biostudies-literature
Publications
A machine-learning heuristic to improve gene score prediction of polygenic traits.
Paré Guillaume G Mao Shihong S Deng Wei Q WQ
Scientific reports 20171004 1
Machine-learning techniques have helped solve a broad range of prediction problems, yet are not widely used to build polygenic risk scores for the prediction of complex traits. We propose a novel heuristic based on machine-learning techniques (GraBLD) to boost the predictive performance of polygenic risk scores. Gradient boosted regression trees were first used to optimize the weights of SNPs included in the score, followed by a novel regional adjustment for linkage disequilibrium. A calibration ...[more]
Similar Datasets
Project description:The popular statistics-based Genome-wide association studies (GWAS) have provided deep insights into the field of complex disorder genetics. However, its clinical applicability to predict disease/trait outcomes remains unclear as statistical models are not designed to make predictions. This study employs statistics-free machine-learning (ML)-optimized polygenic risk score (PRS) to complement existing GWAS and bring the prediction of disease/trait outcomes closer to clinical application. Rheumatoid Arthritis (RA) was selected as a model disease to demonstrate the robustness of ML in disease prediction as RA is a prevalent chronic inflammatory joint disease with high mortality rates, affecting adults at the economic prime. Early identification of at-risk individuals may facilitate measures to mitigate the effects of the disease. This study employs a robust ML feature selection algorithm to identify single nucleotide polymorphisms (SNPs) that can predict RA from a set of training data comprising RA patients and population control samples. Thereafter, selected SNPs were evaluated for their predictive performances across 3 independent, unseen test datasets. The selected SNPs were subsequently used to generate PRS which was also evaluated for its predictive capacity as a sole feature. Through robust ML feature selection, 9 SNPs were found to be the minimum number of features for excellent predictive performance (AUC > 0.9) in 3 independent, unseen test datasets. PRS based on these 9 SNPs was significantly associated with (P < 1 × 10-16) and predictive (AUC > 0.9) of RA in the 3 unseen datasets. A RA ML-PRS calculator of these 9 SNPs was developed ( https://xistance.shinyapps.io/prs-ra/ ) to facilitate individualized clinical applicability. The majority of the predictive SNPs are protective, reside in non-coding regions, and are either predicted to be potentially functional SNPs (pfSNPs) or in high linkage disequilibrium (r2 > 0.8) with un-interrogated pfSNPs. These findings highlight the promise of this ML strategy to identify useful genetic features that can robustly predict disease and amenable to translation for clinical application.
| S-EPMC9903430 | biostudies-literature
Project description:Gene expression profiles were generated from 199 primary breast cancer patients. Samples 1-176 were used in another study, GEO Series GSE22820, and form the training data set in this study. Sample numbers 200-222 form a validation set. This data is used to model a machine learning classifier for Estrogen Receptor Status. RNA was isolated from 199 primary breast cancer patients. A machine learning classifier was built to predict ER status using only three gene features.
2013-01-01 | E-GEOD-29210 | biostudies-arrayexpress
Project description:Polygenic risk scores (PRS) are an emerging tool to predict the clinical phenotypes and outcomes of individuals. Validation and transferability of existing PRS across independent datasets and diverse ancestries are limited, which hinders the practical utility and exacerbates health disparities. We propose PRSmix, a framework that evaluates and leverages the PRS corpus of a target trait to improve prediction accuracy, and PRSmix+, which incorporates genetically correlated traits to better capture the human genetic architecture. We applied PRSmix to 47 and 32 diseases/traits in European and South Asian ancestries, respectively. PRSmix demonstrated a mean prediction accuracy improvement of 1.20-fold (95% CI: [1.10; 1.3]; P-value = 9.17 × 10-5) and 1.19-fold (95% CI: [1.11; 1.27]; P-value = 1.92 × 10-6), and PRSmix+ improved the prediction accuracy by 1.72-fold (95% CI: [1.40; 2.04]; P-value = 7.58 × 10-6) and 1.42-fold (95% CI: [1.25; 1.59]; P-value = 8.01 × 10-7) in European and South Asian ancestries, respectively. Compared to the previously established cross-trait-combination method with scores from pre-defined correlated traits, we demonstrated that our method can improve prediction accuracy for coronary artery disease up to 3.27-fold (95% CI: [2.1; 4.44]; P-value after FDR correction = 2.6 × 10-4). Our method provides a comprehensive framework to benchmark and leverage the combined power of PRS for maximal performance in a desired target population.
| S-EPMC9980241 | biostudies-literature
Project description:In this paper we characterize the performance of linear models trained via widely-used sparse machine learning algorithms. We build polygenic scores and examine performance as a function of training set size, genetic ancestral background, and training method. We show that predictor performance is most strongly dependent on size of training data, with smaller gains from algorithmic improvements. We find that LASSO generally performs as well as the best methods, judged by a variety of metrics. We also investigate performance characteristics of predictors trained on one genetic ancestry group when applied to another. Using LASSO, we develop a novel method for projecting AUC and correlation as a function of data size (i.e., for new biobanks) and characterize the asymptotic limit of performance. Additionally, for LASSO (compressed sensing) we show that performance metrics and predictor sparsity are in agreement with theoretical predictions from the Donoho-Tanner phase transition. Specifically, a future predictor trained in the Taiwan Precision Medicine Initiative for asthma can achieve an AUC of [Formula: see text] and for height a correlation of [Formula: see text] for a Taiwanese population. This is above the measured values of [Formula: see text] and [Formula: see text], respectively, for UK Biobank trained predictors applied to a European population.
| S-EPMC10356957 | biostudies-literature
Project description:Endometriosis is a major health care challenge because many young women with endometriosis go undetected for an extended period, which may lead to pain sensitization. Clinical tools to better identify candidates for laparoscopy-guided diagnosis are urgently needed. Since endometriosis has a strong genetic component, there is a growing interest in using genetics as part of the clinical risk assessment. The aim of this work was to investigate the discriminative ability of a polygenic risk score (PRS) for endometriosis using three different cohorts: surgically confirmed cases from the Western Danish endometriosis referral Center (249 cases, 348 controls), cases identified from the Danish Twin Registry (DTR) based on ICD-10 codes from the National Patient Registry (140 cases, 316 controls), and replication analysis in the UK Biobank (2,967 cases, 256,222 controls). Patients with adenomyosis from the DTR (25 cases) and from the UK Biobank (1,883 cases) were included for comparison. The PRS was derived from 14 genetic variants identified in a published genome-wide association study with more than 17,000 cases. The PRS was associated with endometriosis in surgically confirmed cases [odds ratio (OR) = 1.59, p = 2.57× 10-7] and in cases from the DTR biobank (OR = 1.50, p = 0.0001). Combining the two Danish cohorts, each standard deviation increase in PRS was associated with endometriosis (OR = 1.57, p = 2.5× 10-11), as well as the major subtypes of endometriosis; ovarian (OR = 1.72, p = 6.7× 10-5), infiltrating (OR = 1.66, p = 2.7× 10-9), and peritoneal (OR = 1.51, p = 2.6 × 10-3). These findings were replicated in the UK Biobank with a much larger sample size (OR = 1.28, p < 2.2× 10-16). The PRS was not associated with adenomyosis, suggesting that adenomyosis is not driven by the same genetic risk variants as endometriosis. Our results suggest that a PRS captures an increased risk of all types of endometriosis rather than an increased risk for endometriosis in specific locations. Although the discriminative accuracy is not yet sufficient as a stand-alone clinical utility, our data demonstrate that genetics risk variants in form of a simple PRS may add significant new discriminatory value. We suggest that an endometriosis PRS in combination with classical clinical risk factors and symptoms could be an important step in developing an urgently needed endometriosis risk stratification tool.
| S-EPMC9580817 | biostudies-literature
Project description:The genetic analysis of complex traits does not escape the current excitement around artificial intelligence, including a renewed interest in "deep learning" (DL) techniques such as Multilayer Perceptrons (MLPs) and Convolutional Neural Networks (CNNs). However, the performance of DL for genomic prediction of complex human traits has not been comprehensively tested. To provide an evaluation of MLPs and CNNs, we used data from distantly related white Caucasian individuals (n ∼100k individuals, m ∼500k SNPs, and k = 1000) of the interim release of the UK Biobank. We analyzed a total of five phenotypes: height, bone heel mineral density, body mass index, systolic blood pressure, and waist-hip ratio, with genomic heritabilities ranging from ∼0.20 to 0.70. After hyperparameter optimization using a genetic algorithm, we considered several configurations, from shallow to deep learners, and compared the predictive performance of MLPs and CNNs with that of Bayesian linear regressions across sets of SNPs (from 10k to 50k) that were preselected using single-marker regression analyses. For height, a highly heritable phenotype, all methods performed similarly, although CNNs were slightly but consistently worse. For the rest of the phenotypes, the performance of some CNNs was comparable or slightly better than linear methods. Performance of MLPs was highly dependent on SNP set and phenotype. In all, over the range of traits evaluated in this study, CNN performance was competitive to linear models, but we did not find any case where DL outperformed the linear model by a sizable margin. We suggest that more research is needed to adapt CNN methodology, originally motivated by image analysis, to genetic-based problems in order for CNNs to be competitive with linear models.
| S-EPMC6218236 | biostudies-literature
Project description:Background and objectivesAlzheimer disease (AD) has a polygenic architecture, for which genome-wide association studies (GWAS) have helped elucidate sequence variants (SVs) influencing susceptibility. Polygenic risk score (PRS) approaches show promise for generating summary measures of inherited risk for clinical AD based on the effects of APOE and other GWAS hits. However, existing PRS approaches, based on traditional regression models, explain only modest variation in AD dementia risk and AD-related endophenotypes. We hypothesized that machine learning (ML) models of polygenic risk (ML-PRS) could outperform standard regression-based PRS methods and therefore have the potential for greater clinical utility.MethodsWe analyzed combined data from the Mayo Clinic Study of Aging (n = 1,791) and the Alzheimer's Disease Neuroimaging Initiative (n = 864). An AD PRS was computed for each participant using the top common SVs obtained from a large AD dementia GWAS. In parallel, ML models were trained using those SV genotypes, with amyloid PET burden as the primary outcome. Secondary outcomes included amyloid PET positivity and clinical diagnosis (cognitively unimpaired vs impaired). We compared performance between ML-PRS and standard PRS across 100 training sessions with different data splits. In each session, data were split into 80% training and 20% testing, and then five-fold cross-validation was used within the training set to ensure the best model was produced for testing. We also applied permutation importance techniques to assess which genetic factors contributed most to outcome prediction.ResultsML-PRS models outperformed the AD PRS (r2 = 0.28 vs r2 = 0.24 in test set) in explaining variation in amyloid PET burden. Among ML approaches, methods accounting for nonlinear genetic influences were superior to linear methods. ML-PRS models were also more accurate when predicting amyloid PET positivity (area under the curve [AUC] = 0.80 vs AUC = 0.63) and the presence of cognitive impairment (AUC = 0.75 vs AUC = 0.54) compared with the standard PRS.DiscussionWe found that ML-PRS approaches improved upon standard PRS for prediction of AD endophenotypes, partly related to improved accounting for nonlinear effects of genetic susceptibility alleles. Further adaptations of the ML-PRS framework could help to close the gap of remaining unexplained heritability for AD and therefore facilitate more accurate presymptomatic and early-stage risk stratification for clinical decision-making.
| S-EPMC10798228 | biostudies-literature
Project description:Gout is a painful inflammatory arthritis affecting more than 8 million Americans. Identifying high-risk patients in early life could potentially encourage people to adopt lifestyle changes to prevent gout. Polygenic risk score (PRS) provides an overall estimate of an individual's genetic liability to develop a disease and can be used for early identification of high-risk individuals. In this study, we validated a previously reported PRS in an independent cohort. The urate-PRS was constructed from 110 significant urate-associated variants identified in Europeans. Phenome-wide and PRS-wide association study showed the urate-PRS is highly specifically associated with gout (phecode: 274.10; beta = 1.495 [1.372, 1.619], p = 4.37e-124). Urate-PRS alone did not performed in the gout prediction (area under the receiver operating characteristic curve, AUROC = 0.640); however, the addition of PRS upon demographics significantly improved the model performance, yielding an AUROC of 0.804 from 0.777 (DeLong test p = 3.66e-9). Trans-ethnic PRS and European-specific PRS showed similar prediction performance. We observed increasing gout prevalence and odds ratio (OR) across the PRS quintiles. Our study showed 8.2% of the cohort had more than 2.5 odds for gout than remainders, indicating that urate-PRS may be a better marker than age and sex to stratify patient risk. With the rapid growth of large biorepositories, such as All of Us, urate-PRS can be applied quickly and widely in population to estimate individual's risk, providing a powerful tool for gout preventive purpose in population health.
| S-EPMC7889590 | biostudies-literature
Project description:A primary goal of polygenic scores, which aggregate the effects of thousands of trait-associated DNA variants discovered in genome-wide association studies (GWASs), is to estimate individual-specific genetic propensities and predict outcomes. This is typically achieved using a single polygenic score, but here we use a multi-polygenic score (MPS) approach to increase predictive power by exploiting the joint power of multiple discovery GWASs, without assumptions about the relationships among predictors. We used summary statistics of 81 well-powered GWASs of cognitive, medical and anthropometric traits to predict three core developmental outcomes in our independent target sample: educational achievement, body mass index (BMI) and general cognitive ability. We used regularized regression with repeated cross-validation to select from and estimate contributions of 81 polygenic scores in a UK representative sample of 6710 unrelated adolescents. The MPS approach predicted 10.9% variance in educational achievement, 4.8% in general cognitive ability and 5.4% in BMI in an independent test set, predicting 1.1%, 1.1%, and 1.6% more variance than the best single-score predictions. As other relevant GWA analyses are reported, they can be incorporated in MPS models to maximize phenotype prediction. The MPS approach should be useful in research with modest sample sizes to investigate developmental, multivariate and gene-environment interplay issues and, eventually, in clinical settings to predict and prevent problems using personalized interventions.
| S-EPMC5681246 | biostudies-literature
Project description:Polygenic risk score (PRS) is useful for capturing an individual's genetic susceptibility. However, previous studies have not fully exploited the potential of the risk factor PRS (RFPRS) for disease prediction. We explored the potential of integrating disease-related RFPRSs with disease PRS to enhance disease prediction performance. We constructed 112 RFPRSs and analyzed the association of RFPRSs with diseases to identify disease-related RFPRSs in 700 diseases, using the UK Biobank dataset. We uncovered 6157 statistically significant associations between 247 diseases and 109 RFPRSs. We estimated the disease PRSs of 70 diseases that exhibited statistically significant heritability, to generate RFDiseasemetaPRS-a combined PRS integrating RFPRSs and disease PRS-and compare the prediction performance metrics between RFDiseasemetaPRS and disease PRS. RFDiseasemetaPRS showed better performance for Nagelkerke's pseudo-R2, odds ratio (OR) per 1 SD, net reclassification improvement (NRI) values and difference of R2 considered by variance of R2 in 31 out of 70 diseases. Additionally, we assessed risk classification between two models by examining OR between the top 10% and remaining 90% individuals for the 31 diseases; RFDiseasemetaPRS exhibited better R2, NRI and OR than disease PRS. These findings highlight the importance of utilizing RFDiseasemetaPRS, which can provide personalized healthcare and tailored prevention strategies.
| S-EPMC10864389 | biostudies-literature