Regularized Machine Learning in the Genetic Prediction of Complex Traits
Open accessThe DOI matches its Crossref record; 18 of the 18 references it cites were found in a registry or an index.
Abstract
OverviewCompared to univariate analysis of genome-wide association (GWA) studies, machine learning-based models have been shown to provide improved means of learning such multilocus panels of genetic variants and their interactions that are most predictive of complex phenotypic traits.Many applications of predictive modeling rely on effective variable selection, often implemented through model regularization, which penalizes the model complexity and enables predictions in individuals outside of the training dataset.However, the different regularization approaches may also lead to considerable differences, especially in the number of genetic variants needed for maximal predictive accuracy, as illustrated here in examples from both disease classification and quantitative trait prediction.We also highlight the potential pitfalls of the regularized machine learning models, related to issues such as model overfitting to the training data, which may lead to overoptimistic prediction results, as well as identifiability of the predictive variants, which is important in many medical applications.While genetic risk prediction for human diseases is used as a motivating use case, we argue that these models are also widely applicable in nonhuman applications, such as animal and plant breeding, where accurate genotype-to-phenotype modeling is needed.Finally, we discuss some key future advances, open questions and challenges in this developing field, when moving toward low-frequency variants and cross-phenotype interactions.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Breiman, L. (2001) · Machine Learning · doi:10.1023/a:1010933404324 · Cited by 133,538
Regularization and Variable Selection Via the Elastic Net
Zou, H., & Hastie, T. (2005) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.1467-9868.2005.00503.x · Cited by 20,957
Common SNPs explain a large proportion of the heritability for human height
Yang, J., Benyamin, B., McEvoy, B. P., Gordon, S., Henders, A. K., Nyholt, D. R., Madden, P. A., Heath, A. C., Martin, N. G., Montgomery, G. W., et al. (2010) · Nature Genetics · doi:10.1038/ng.608 · Cited by 4,446
Theoretical and Empirical Analysis of ReliefF and RReliefF
Robnik-Šikonja, M., & Kononenko, I. (2003) · Machine Learning · doi:10.1023/a:1025667309714 · Cited by 3,025 · cited 1×
“Whereas classical, univariate filter methods evaluate the relevance of each genetic feature independently of the others, more advanced multivariate filters have also been proposed, including the Relief family of approaches [101].”
MethodPersonal genomes: The case of the missing heritability
Maher, B. (2008) · Nature · doi:10.1038/456018a · Cited by 1,737 · cited 1×
“While rare variants have been proposed as one explanation for the missing heritability [36,37], there has been a divergence of opinion over whether rare variants of large effect or common variants of small effect are contributing most to the phenotypic variability [38].”
BackgroundPower and predictive accuracy of polygenic risk scores
Dudbridge, F. (2013) · PLoS Genetics · doi:10.1371/journal.pgen.1003348 · Cited by 1,509 · cited 3×
“The study also confirmed the projections from previous works [31–33], suggesting that predictive accuracy is highly dependent on the sample sizes and the spectrum of variants included in the model, in addition to the heritability of the disease trait.”
BackgroundResultDetecting gene–gene interactions that underlie human diseases
Cordell, H. J. (2009) · Nature Reviews Genetics · doi:10.1038/nrg2579 · Cited by 1,350 · cited 1×
“Machine learning models can also deal with genetic interactions, which are known to play an important role in the development and treatment of many complex diseases [12–16], but are often missed by single-locus association tests [17].”
BackgroundRare and common variants: twenty arguments
Gibson, G. (2012) · Nature Reviews Genetics · doi:10.1038/nrg3118 · Cited by 1,278 · cited 1×
“While rare variants have been proposed as one explanation for the missing heritability [36,37], there has been a divergence of opinion over whether rare variants of large effect or common variants of small effect are contributing most to the phenotypic variability [38].”
BackgroundChapter 11: Genome-wide association studies
Bush, W. S., & Moore, J. H. (2012) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1002822 · Cited by 1,171
Advantages and pitfalls in the application of mixed-model association methods
Yang, J., Zaitlen, N. A., Goddard, M. E., Visscher, P. M., & Price, A. L. (2014) · Nature Genetics · doi:10.1038/ng.2876 · Cited by 981 · cited 1×
“Lasso has been extended to account for population structures through linear mixed models [105], which are gaining much popularity in association studies [106].”
BackgroundGenome-wide association analysis by lasso penalized logistic regression
Wu, T. T., Chen, Y. F., Hastie, T., Sobel, E., & Lange, K. (2009) · Bioinformatics · doi:10.1093/bioinformatics/btp041 · Cited by 785 · cited 1×
“There are also other implementations available, including Mendel [85], HyperLasso [86] and SparSNP [87], gpu-lasso [88], and PUMA [89].”
MethodClinical assessment incorporating a personal genome
Ashley, E. A., Butte, A. J., Wheeler, M. T., Chen, R., Klein, T. E., Dewey, F. E., Dudley, J. T., Ormond, K. E., Pavlovic, A., Morgan, A. A., et al. (2010) · The Lancet · doi:10.1016/s0140-6736(10)60452-7 · Cited by 674
Ripatti, S., Tikkanen, E., Orho-Melander, M., Havulinna, A. S., Silander, K., Sharma, A., Guiducci, C., Perola, M., Jula, A., Sinisalo, J., et al. (2010) · The Lancet · doi:10.1016/s0140-6736(10)61267-6 · Cited by 562
Self-citationIntegrative approaches for finding modular structure in biological networks
Mitra, K., Carvunis, A.-R., Ramesh, S. K., & Ideker, T. (2013) · Nature Reviews Genetics · doi:10.1038/nrg3552 · Cited by 535 · cited 1×
“Networks of genetic and/or physical interactions may therefore serve as useful prior information for the prediction models to move from variant-level features towards pathway-level features [5,73–75].”
BackgroundSimultaneous Analysis of All SNPs in Genome-Wide and Re-Sequencing Association Studies
Hoggart, C. J., Whittaker, J. C., De Iorio, M., & Balding, D. J. (2008) · PLoS Genetics · doi:10.1371/journal.pgen.1000130 · Cited by 321 · cited 1×
“There are also other implementations available, including Mendel [85], HyperLasso [86] and SparSNP [87], gpu-lasso [88], and PUMA [89].”
MethodHints of hidden heritability in GWAS
Gibson, G. (2010) · Nature Genetics · doi:10.1038/ng0710-558 · Cited by 281
Association screening of common and rare genetic variants by penalized regression
Zhou, H., Sehl, M. E., Sinsheimer, J. S., & Lange, K. (2010) · Bioinformatics · doi:10.1093/bioinformatics/btq448 · Cited by 137 · cited 1×
“Regularization methods based on sparse group Lasso, for instance, can be extended to rare variants and pathway-driven variant selection [22,43].”
MethodAccommodating Linkage Disequilibrium in Genetic-Association Analyses via Ridge Regression
Malo, N., Libiger, O., & Schork, N. J. (2008) · The American Journal of Human Genetics · doi:10.1016/j.ajhg.2007.10.012 · Cited by 133 · cited 1×
“Penalization strategies, such as ridge regression, Lasso and RLS, allow the model to avoid placing too much weight on potentially overfit variables in the presence of LD, which can lead to improved selection of causal variants [110,111].”
Method
Cited by
Analysis of polygenic risk score usage and performance in diverse human populations
Duncan, L., Shen, H., Gelaye, B., Meijsen, J., Ressler, K., Feldman, M., Peterson, R., & Domingue, B. (2019) · Nature Communications · doi:10.1038/s41467-019-11112-0 · Cited by 1,038
A Review of Feature Selection Methods for Machine Learning-Based Disease Risk Prediction
Pudjihartono, N., Fadason, T., Kempa-Liehr, A. W., & O'Sullivan, J. M. (2022) · Frontiers in Bioinformatics · doi:10.3389/fbinf.2022.927312 · Cited by 880
Psychiatric genetics and the structure of psychopathology
Smoller, J. W., Andreassen, O. A., Edenberg, H. J., Faraone, S. V., Glatt, S. J., & Kendler, K. S. (2019) · Molecular Psychiatry · doi:10.1038/s41380-017-0010-4 · Cited by 398
“, [69]) have been intriguing, but their inter-”
BackgroundNaulaerts, S., Dang, C. C., & Ballester, P. J. (2017) · Oncotarget · doi:10.18632/oncotarget.20923 · Cited by 41
“Beyond this research area, multi-variate machine-learning models are also starting to be advocated for genomicbased prediction of other complex phenotypic traits [27].”
BackgroundOn to the next chapter for crop breeding: Convergence with data science
Ersoz, E. S., Martin, N. F., & Stapleton, A. E. (2020) · Crop Science · doi:10.1002/csc2.20054 · Cited by 33
Metadata
- Authors
- Okser, S., Pahikkala, T., Airola, A., Salakoski, T., Ripatti, S., & Aittokallio, T.
- Venue
- PLoS Genetics
- Publisher
- Public Library of Science (PLoS)
- Published
- 2014-11-13
- Volume
- 10
- Issue
- 11
- Pages
- e1004754
- ISSN
- 1553-7390 (ISSN-L)
- 1553-7404
- OpenAlex
- W1964866242
- Cited by
- 156
- References
- 18
- FWCI
- 5.55
- Type
- journal article
- Language
- en
- Open access
- gold
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.