Comprehensive Assessment of Genotype Imputation Performance

Shi, Shuo; Ni, Yuan; Yang, Ming; Du, Zhenglin; Wang, Jinyue; Sheng, Xin; Wu, Jiayan; Xiao, Jingfa

doi:10.1159/000489758

Cited by 59 publications

(51 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(To see this, consider the case where c = tk , with k a positive integer, and place key sites at t/ 2 , t , 3 t/ 2 ,…, c − t/ 2.) This means that with a minimum reported IBS threshold of 1cM, 100 uploaded datasets could reveal approximately 100 genotypes per cM, which is enough to impute genome-wide genotypes at 97 − 98% accuracy (Shi et al, 2018). In principle, the key sites could also be chosen to ensure good LD coverage and higher imputation accuracy.…”

Section: Resultsmentioning

confidence: 99%

Attacks on genetic privacy via uploads to genealogical databases

Edge

Coop

2019

Preprint

View full text Add to dashboard Cite

7Direct-to-consumer (DTC) genetics services are increasingly popular for genetic genealogy, with 8 tens of millions of customers as of 2019. Several DTC genealogy services allow users to upload 9 their own genetic datasets in order to search for genetic relatives. A user and a target person in 10 the database are identified as genetic relatives if the user's uploaded genome shares one or more 11 sufficiently long segments in common with that of the target person-that is, if the two genomes 12 share one or more long regions identical by state (IBS). IBS matches reveal some information 13 about the genotypes of the target person, particularly if the chromosomal locations of IBS matches 14 are shared with the uploader. Here, we describe several methods by which an adversary who 15 wants to learn the genotypes of people in the database can do so by uploading multiple datasets. 16Depending on the methods used for IBS matching and the information about IBS segments 17 returned to the user, substantial information about users' genotypes can be revealed with a few 18 hundred uploaded datasets. For example, using a method we call IBS tiling, we estimate that an 19 adversary who uploads approximately 900 publicly available genomes could recover at least one 20 allele at SNP sites across up to 82% of the genome of a median person of European ancestries. 21In databases that detect IBS segments using unphased genotypes, approximately 100 uploads of 22 falsified datasets can reveal enough genetic information to allow accurate genome-wide imputation 23 of every person in the database. We provide simple-to-implement suggestions that will prevent the 24 exploits we describe and discuss our results in light of recent trends in genetic privacy, including 25 the recent use of uploads to DTC genetic genealogy services by law enforcement. 26(Regalado, 2019). One of the major applications of DTC genetics has been genetic genealogy. 31Customers of companies such as 23andMe and Ancestry, once they are genotyped, can view a list 32 of other customers who are likely to be genetic relatives. These putative relatives' full names are 33 often given, and sometimes contact details are given as well. Such genealogical matching services 34 are of interest to people who want to find distant genetic relatives to extend their family tree, and 35 can be particularly useful to people who otherwise may not have information about their genetic 36 relatives, such as adoptees or the biological children of sperm donors. Several genetic genealogy 37 services-including GEDmatch, MyHeritage, FamilyTreeDNA, and LivingDNA (Table 1)-allow 38 users to upload their own genetic data if they have been genotyped by another company. These 39 entities generally offer some subset of their services at no charge to uploaders, which helps to 40 grow their databases. Upload services have also been used by law enforcement, with the goal of 41 identifying relatives of the source of a crime-scene sample (Erlich et al., 2018; Edge and Coop, 42 2019), prompting discussion abo...

show abstract

Section: Resultsmentioning

confidence: 99%

Attacks on genetic privacy via uploads to genealogical databases

Edge

Coop

2019

Preprint

View full text Add to dashboard Cite

show abstract

“…The empirical performance of imputation methods has been extensively studied [3,23,17,4], while the theoretical behavior and limit of the underlying framework has not been well characterized, especially for rare variants. We formulate this problem of imputation accuracy into a coalescent model, considering imputing the missing genotypes on one target haplotype by copying from a set of reference haplotypes that share the most recent common ancestor (MRCA) with the target.…”

Section: Discussionmentioning

confidence: 99%

Why are rare variants hard to impute? Coalescent models reveal theoretical limits in existing algorithms

Zoellner

2020

Preprint

View full text Add to dashboard Cite

Genotype imputation is an indispensable step in human genetic studies. Large reference panels with deeply sequenced genomes now allow interrogating variants with minor allele frequency < 1% without sequencing. While it is critical to consider limits of this approach, imputation methods for rare variants have only done so empirically; the theoretical basis of their imputation accuracy has not been explored. To provide theoretical consideration of imputation accuracy under the current imputation framework, we develop a coalescent model of imputing rare variants, leveraging the joint genealogy of the sample to be imputed and reference individuals. We show that broadly used imputation algorithms includes model miss-specifications about this joint genealogy that limit the ability to correctly impute rare variants. We develop closed-form solutions for the probability distribution of this joint genealogy and quantify the inevitable error rate resulting from the model miss-specification across a range of allele frequencies and reference sample sizes. We show that the probability of a falsely imputed minor allele decreases with reference sample size, but the proportion of falsely imputed minor alleles mostly depends on the allele count in the reference sample. We summarize the impact of this error on genotype imputation on association tests by calculating the r2 between imputed and true genotype and show that even when modeling other sources of error, the impact of the model miss-specification have a significant impact on the r2 of rare variants. These results provide a framework for developing new imputation algorithms and for interpreting rare variant association analyses.

show abstract

“…Variability of the imputation could have been caused by presence of the rare alleles in the untyped or imputed SNPs [39,40]. SNP density in the region which could have slightly lowered the accuracy score [41]. As a result, there were 1013 SNPs in total after imputation.…”

Section: Snp Imputationmentioning

confidence: 99%

Association of PARP1-specific polymorphisms and haplotypes with non-small cell lung cancer subtypes

et al. 2020

View full text Add to dashboard Cite

Objective The carcinogenesis role of PARP1 in lung cancer is still not clear. Analysis at allelic levels cannot fully explain the function of PARP1 on lung cancer. Our study aims to further explore the relation between PARP1 haplotypes and lung cancer. Materials and methods DNA and RNA were extracted from non-small cell lung cancer (NSCLC) tumor and adjacent normal fresh frozen tissue. Five PARP1-SNPs were genotyped and PARP1-specific SNPs were imputed using IMPUTE and SHAPEIT software. The SNPs were subjected to allelic, haplotype and SNP-SNP interaction analyses. Correlation between SNPs and mRNA/protein expressions were performed. Results SNP imputation inferred the ungenotyped SNPs and increased the power for association analysis. Tumor tissue samples are more likely to carry rs1805414 (OR = 1.85; 95% CI: 1.12–3.06; P-value: 0.017) and rs1805404 (OR = 2.74; 95%CI 1.19–6.32; P-value: 0.015) compared to normal tissues. Our study is the first study to show that haplotypes comprising of 5 SNPs on PARP1 (rs1136410, rs3219073, rs1805414, rs1805404, rs1805415) is able to differentiate the NSCLC tumor from normal tissues. Interaction between rs3219073, rs1805415, and rs1805414 were significantly associated with the NSCLC tumor with OR ranging from 3.61–6.75; 95%CI from 1.82 to 19.9; P-value<0.001. Conclusion PARP1 haplotypes may serve as a better predictor in lung cancer development and prognosis compared to single alleles.

show abstract

Comprehensive Assessment of Genotype Imputation Performance

Cited by 59 publications

References 30 publications

Attacks on genetic privacy via uploads to genealogical databases

Attacks on genetic privacy via uploads to genealogical databases

Why are rare variants hard to impute? Coalescent models reveal theoretical limits in existing algorithms

Association of PARP1-specific polymorphisms and haplotypes with non-small cell lung cancer subtypes

Contact Info

Product

Resources

About