Detecting gene-gene interactions using a permutation-based random forest method

Li, Jing; Malley, James D.; Andrew, Angeline S.; Karagas, Margaret R.; Moore, Jason H.

doi:10.1186/s13040-016-0093-5

Cited by 61 publications

(45 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, this is an area of active research, and the capacity of RF models in their current form to both capture and identify SNP interactions has been disputed (Winham et al., ; Wright et al., ). New modifications of RF models are being developed to more effectively identify interaction effects (e.g., Li, Malley, Andrew, Karagas, & Moore, ), but these models are computationally demanding and are not designed for large data sets. Overall, extensions of RF show potential for identifying more complex genetic architectures on small sets of loci, but caution is warranted in using them on empirical data prior to rigorous testing on realistic simulation scenarios.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Comparing methods for detecting multilocus adaptation with multivariate genotype–environment associations

et al. 2018

View full text Add to dashboard Cite

Identifying adaptive loci can provide insight into the mechanisms underlying local adaptation. Genotype-environment association (GEA) methods, which identify these loci based on correlations between genetic and environmental data, are particularly promising. Univariate methods have dominated GEA, despite the high dimensional nature of genotype and environment. Multivariate methods, which analyse many loci simultaneously, may be better suited to these data as they consider how sets of markers covary in response to environment. These methods may also be more effective at detecting adaptive processes that result in weak, multilocus signatures. Here, we evaluate four multivariate methods and five univariate and differentiation-based approaches, using published simulations of multilocus selection. We found that Random Forest performed poorly for GEA. Univariate GEAs performed better, but had low detection rates for loci under weak selection. Constrained ordinations, particularly redundancy analysis (RDA), showed a superior combination of low false-positive and high true-positive rates across all levels of selection. These results were robust across the demographic histories, sampling designs, sample sizes and weak population structure tested here. The value of combining detections from different methods was variable and depended on the study goals and knowledge of the drivers of selection. Re-analysis of genomic data from grey wolves highlighted the unique, covarying sets of adaptive loci that could be identified using RDA. Although additional testing is needed, this study indicates that RDA is an effective means of detecting adaptation, including signatures of weak, multilocus selection, providing a powerful tool for investigating the genetic basis of local adaptation.

show abstract

Section: Discussionmentioning

confidence: 99%

“…Habitat and X-coordinate variables Wright et al, 2016). New modifications of RF models are being developed to more effectively identify interaction effects (e.g., Li, Malley, Andrew, Karagas, & Moore, 2016), but these models are computationally demanding and are not designed for large data sets.…”

Section: Habitat Variable Onlymentioning

confidence: 99%

Comparing methods for detecting multilocus adaptation with multivariate genotype–environment associations

et al. 2018

View full text Add to dashboard Cite

show abstract

“…For investigations of associations with leaf shape in large‐scale genome‐wide association studies, refer to Li et al . () and Winham et al . () for possible limitations and improvements when applying RF to detect interactions.…”

Section: Discussionmentioning

confidence: 99%

“…Our methods could be applicable to other shape data collected from common gardens or growth chambers, with different genotyping technologies, more variables, a higher density of genetic markers, or more samples. For investigations of associations with leaf shape in large-scale genome-wide association studies, refer to Li et al (2016) and Winham et al (2012) for possible limitations and improvements when applying RF to detect interactions.…”

Section: Researchmentioning

confidence: 99%

Quantitative gene–gene and gene–environment mapping for leaf shape variation using tree‐based models

Dai

Symanzik

et al. 2016

New Phytologist

View full text Add to dashboard Cite

Leaf shape traits have long been a focus of many disciplines, but the complex genetic and environmental interactive mechanisms regulating leaf shape variation have not yet been investigated in detail. The question of the respective roles of genes and environment and how they interact to modulate leaf shape is a thorny evolutionary problem, and sophisticated methodology is needed to address it. In this study, we investigated a framework-level approach that inputs shape image photographs and genetic and environmental data, and then outputs the relative importance ranks of all variables after integrating shape feature extraction, dimension reduction, and tree-based statistical models. The power of the proposed framework was confirmed by simulation and a Populus szechuanica var. tibetica data set. This new methodology resulted in the detection of novel shape characteristics, and also confirmed some previous findings. The quantitative modeling of a combination of polygenetic, plastic, epistatic, and gene-environment interactive effects, as investigated in this study, will improve the discernment of quantitative leaf shape characteristics, and the methods are ready to be applied to other leaf morphology data sets. Unlike the majority of approaches in the quantitative leaf shape literature, this framework-level approach is data-driven, without assuming any pre-known shape attributes, landmarks, or model structures.

show abstract

“…MDR is superior to traditional statistical methods in case-control studies for analyzing gene-gene interactions. This method is convenient for users to assess interactions because MDR is model-free, where no genetic pattern is supposed, and uses known data about the disease for which inheritance-model is unknown or are extreme complex (Li et al, 2016). This software can be freely obtained from the Internet (http://sourceforge.net/projects/mdr/).…”

Section: Discussionmentioning

confidence: 99%

Gene-gene interaction between VANGL1, FZD3, and FZD6 correlated with neural tube defects in Han population of Northern China

Zhang¹,

Fang²,

Wu³

et al. 2016

Genet. Mol. Res.

View full text Add to dashboard Cite

ABSTRACT. We evaluated the influence of gene-gene interactions between VANGL1, FZD3, and FZD6 on the risk of neural tube defects (NTDs) in Han population in the north of China. Two single nucleotide polymorphisms (SNPs) (rs4839469 and rs34059106) within VANGL1, two SNPs (rs2241802 and rs28639533) within FZD3, and three SNPs (rs827528, rs3808553, and rs12549394) within FZD6 were genotyped in 135 NTD patients and 135 controls. The gene-gene 2 R.P. Zhang et al. Genetics and Molecular Research 15 (3): gmr.15039010 interactions between VANGL1, FZD3, and FZD6 were analyzed using multifactor dimensionality reduction (MDR) software. The distribution of genotypes of rs4 839 469 within VANGL1 and rs3 808 553 within FZD6 differed significantly difference between patients and controls (P < 0.05). MDR revealed significance in models with 2 SNPs (rs4839469 and rs3808553) (OR = 3.18, 95%CI = 1.85-5.44; χ 2 = 18.39, P < 0.0001), 3 SNPs (rs4839469, rs2241802, and rs3808553) (OR = 4.17, 95%CI = 2.43-7.14; χ 2 = 28.5, P < 0.0001), and 4 SNPs (rs4839469, rs2241802, rs827528, and rs3808553) (OR = 7.34, 95%CI = 3.98-13.54; χ 2 = 45.3, P < 0.0001). Gene-gene interaction between VANGL1, FZD3, and FZD6 may exist, which may increase the risk of NTDs. This work provides insight into the understanding of the etiology of NTDs.

show abstract

Detecting gene-gene interactions using a permutation-based random forest method

Cited by 61 publications

References 37 publications

Comparing methods for detecting multilocus adaptation with multivariate genotype–environment associations

Comparing methods for detecting multilocus adaptation with multivariate genotype–environment associations

Quantitative gene–gene and gene–environment mapping for leaf shape variation using tree‐based models

Gene-gene interaction between VANGL1, FZD3, and FZD6 correlated with neural tube defects in Han population of Northern China

Contact Info

Product

Resources

About