Epistatic Features and Machine Learning Improve Alzheimer's Risk Prediction Over Polygenic Risk Scores
Polygenic risk scores (PRS) are linear combinations of genetic markers weighted by effect size that are commonly used to predict disease risk. For complex heritable diseases such as late onset Alzheimer's disease (LOAD), PRS models fail to capture much of the heritability. Additionally, PRS mod...
Saved in:
Published in | medRxiv : the preprint server for health sciences |
---|---|
Main Authors | , , , , , , , |
Format | Journal Article |
Language | English |
Published |
United States
15.03.2023
|
Online Access | Get more information |
Cover
Loading…
Abstract | Polygenic risk scores (PRS) are linear combinations of genetic markers weighted by effect size that are commonly used to predict disease risk. For complex heritable diseases such as late onset Alzheimer's disease (LOAD), PRS models fail to capture much of the heritability. Additionally, PRS models are highly dependent on the population structure of data on which effect sizes are assessed, and have poor generalizability to new data.
The goal of this study is to construct a
that, in addition to single genetic marker data used in PRS, incorporates epistatic interaction features and machine learning methods to predict lifetime risk for LOAD.
We construct a new state-of-the-art genetic model for lifetime risk of Alzheimer's disease. Our approach innovates over PRS models in two ways: First, by directly incorporating epistatic interactions between SNP loci using an evolutionary algorithm guided by shared pathway information; and second, by estimating risk via an ensemble of machine learning models (gradient boosting machines and deep learning) instead of simple logistic regression. We compare the paragenic model to a PRS model from the literature trained on the same dataset.
The paragenic model is significantly more accurate than the PRS model under 10-fold cross-validation, obtaining an AUC of 83% and near-clinically significant matched sensitivity/specificity of 75%, and remains significantly more accurate when evaluated on an independent holdout dataset. Additionally, the paragenic model maintains accuracy within
genotypes.
Paragenic models show potential for improving lifetime disease risk prediction for complex heritable diseases such as LOAD over PRS models. |
---|---|
AbstractList | Polygenic risk scores (PRS) are linear combinations of genetic markers weighted by effect size that are commonly used to predict disease risk. For complex heritable diseases such as late onset Alzheimer's disease (LOAD), PRS models fail to capture much of the heritability. Additionally, PRS models are highly dependent on the population structure of data on which effect sizes are assessed, and have poor generalizability to new data.
The goal of this study is to construct a
that, in addition to single genetic marker data used in PRS, incorporates epistatic interaction features and machine learning methods to predict lifetime risk for LOAD.
We construct a new state-of-the-art genetic model for lifetime risk of Alzheimer's disease. Our approach innovates over PRS models in two ways: First, by directly incorporating epistatic interactions between SNP loci using an evolutionary algorithm guided by shared pathway information; and second, by estimating risk via an ensemble of machine learning models (gradient boosting machines and deep learning) instead of simple logistic regression. We compare the paragenic model to a PRS model from the literature trained on the same dataset.
The paragenic model is significantly more accurate than the PRS model under 10-fold cross-validation, obtaining an AUC of 83% and near-clinically significant matched sensitivity/specificity of 75%, and remains significantly more accurate when evaluated on an independent holdout dataset. Additionally, the paragenic model maintains accuracy within
genotypes.
Paragenic models show potential for improving lifetime disease risk prediction for complex heritable diseases such as LOAD over PRS models. |
Author | Cruchaga, Carlos Hermes, Stephen Armentrout, Steven Greytak, Ellen McRae Cady, Janet O'Connor, James Carlson, Sarah Wingo, Thomas |
Author_xml | – sequence: 1 givenname: Stephen surname: Hermes fullname: Hermes, Stephen organization: Parabon NanoLabs, Inc., Reston, Virginia, USA – sequence: 2 givenname: Janet surname: Cady fullname: Cady, Janet organization: Parabon NanoLabs, Inc., Reston, Virginia, USA – sequence: 3 givenname: Steven surname: Armentrout fullname: Armentrout, Steven organization: Parabon NanoLabs, Inc., Reston, Virginia, USA – sequence: 4 givenname: James surname: O'Connor fullname: O'Connor, James organization: Parabon NanoLabs, Inc., Reston, Virginia, USA – sequence: 5 givenname: Sarah surname: Carlson fullname: Carlson, Sarah organization: Parabon NanoLabs, Inc., Reston, Virginia, USA – sequence: 6 givenname: Carlos surname: Cruchaga fullname: Cruchaga, Carlos organization: Hope Center Program on Protein Aggregation and Neurodegeneration, Washington University St. Louis, MO, USA – sequence: 7 givenname: Thomas surname: Wingo fullname: Wingo, Thomas organization: Department of Human Genetics, Emory University School of Medicine, Atlanta, GA, USA – sequence: 8 givenname: Ellen McRae surname: Greytak fullname: Greytak, Ellen McRae organization: Parabon NanoLabs, Inc., Reston, Virginia, USA |
BackLink | https://www.ncbi.nlm.nih.gov/pubmed/36798198$$D View this record in MEDLINE/PubMed |
BookMark | eNqFjrsKwkAQRbdQfP-CTGdloYImpUiCgqKofVg3k2Qw-2B3E4hfbxCtrW5xDoc7ZB2lFfZYf7XehMEiDAYsiww5zz0JiJH7yqIDrlI4cVGQQjgit4pUDgdprK4RtuWrQJJoZw6u5J5wsZiS8KQVnGu0cNFlk6Nqgx98E7ptjlk346XDyXdHbBpH991-bqqHxDQxliS3TfI7tvwrvAFcxkGD |
ContentType | Journal Article |
CorporateAuthor | Alzheimer’s Disease Neuroimaging Initiative |
CorporateAuthor_xml | – name: Alzheimer’s Disease Neuroimaging Initiative |
DBID | NPM |
DatabaseName | PubMed |
DatabaseTitle | PubMed |
DatabaseTitleList | PubMed |
Database_xml | – sequence: 1 dbid: NPM name: PubMed url: https://proxy.k.utb.cz/login?url=http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed sourceTypes: Index Database |
DeliveryMethod | no_fulltext_linktorsrc |
ExternalDocumentID | 36798198 |
Genre | Preprint |
GrantInformation_xml | – fundername: NIA NIH HHS grantid: P30 AG072973 – fundername: NIA NIH HHS grantid: P30 AG062421 – fundername: NIA NIH HHS grantid: U24 AG072122 – fundername: NIA NIH HHS grantid: P30 AG066508 – fundername: NIA NIH HHS grantid: P30 AG062429 – fundername: NIA NIH HHS grantid: P20 AG068024 – fundername: NIA NIH HHS grantid: U01 AG024904 – fundername: NIA NIH HHS grantid: RF1 AG058501 – fundername: NIA NIH HHS grantid: P30 AG072947 – fundername: NIA NIH HHS grantid: P30 AG072976 |
GroupedDBID | NPM |
ID | FETCH-pubmed_primary_367981982 |
IngestDate | Sun Oct 13 10:32:04 EDT 2024 |
IsPeerReviewed | false |
IsScholarly | false |
Language | English |
LinkModel | OpenURL |
MergedId | FETCHMERGED-pubmed_primary_367981982 |
PMID | 36798198 |
ParticipantIDs | pubmed_primary_36798198 |
PublicationCentury | 2000 |
PublicationDate | 2023-Mar-15 |
PublicationDateYYYYMMDD | 2023-03-15 |
PublicationDate_xml | – month: 03 year: 2023 text: 2023-Mar-15 day: 15 |
PublicationDecade | 2020 |
PublicationPlace | United States |
PublicationPlace_xml | – name: United States |
PublicationTitle | medRxiv : the preprint server for health sciences |
PublicationTitleAlternate | medRxiv |
PublicationYear | 2023 |
Score | 3.7599418 |
Snippet | Polygenic risk scores (PRS) are linear combinations of genetic markers weighted by effect size that are commonly used to predict disease risk. For complex... |
SourceID | pubmed |
SourceType | Index Database |
Title | Epistatic Features and Machine Learning Improve Alzheimer's Risk Prediction Over Polygenic Risk Scores |
URI | https://www.ncbi.nlm.nih.gov/pubmed/36798198 |
hasFullText | |
inHoldings | 1 |
isFullTextHit | |
isPrint | |
link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnZ3fS8MwEMcPpyB7EcXfP0YeBB_KBNetnY9DJmOwOeaEvY10TXSgXbHd0P313uVHO4YT9aWMHE3bfdrkknzvAnApuS_DMe0RJkWAA5TwtszrPCwLn_vcrdZDqdIXd7pe66naHtaG-f6dKrokDa7Hi2_jSv5DFcuQK0XJ_oFsVikW4G_ki0ckjMdfMW7G5P1RylXy5GY4cja6CRJICps79dnRMwfCabwuXsRE7ZfiJ06fVOW9d1qpUS_BAz4_6eE-8XokryfzI2W5TJY9WOw9-x-TuWMFITGlxZxEqUPTu1gByRZ1cKVjetfMa29hL6BbJaMtyxdAdFPf5lEej42vIM08T2epOWOen2AlOnquQSl9l6cvKi7pt3QA5xKg-E0RcmlJ6EbvSv2zdSVHtjUVoODXqZ3r9jpF2LbFK0MG5ToMdmHH-PysoQHuwYaI9kFm8JiFxxAeM_CYhccMPJbBu0oYsWE5OkboWIZOmzW6AyjdNwd3rbK-tVGss4uM7E1XDmEzmkbiGJhwBZce97gIQnRvPT7m6Hajc8YD0ru7J3C0ppLTtZYzKOY4zmFL4jchLtDBSoOS-v--AKeNNxo |
link.rule.ids | 786 |
linkProvider | National Library of Medicine |
openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Epistatic+Features+and+Machine+Learning+Improve+Alzheimer%27s+Risk+Prediction+Over+Polygenic+Risk+Scores&rft.jtitle=medRxiv+%3A+the+preprint+server+for+health+sciences&rft.au=Hermes%2C+Stephen&rft.au=Cady%2C+Janet&rft.au=Armentrout%2C+Steven&rft.au=O%27Connor%2C+James&rft.date=2023-03-15&rft_id=info%3Apmid%2F36798198&rft_id=info%3Apmid%2F36798198&rft.externalDocID=36798198 |