LFMM 2: Fast and Accurate Inference of Gene-Environment Associations in Genome-Wide Studies

Abstract

Gene-environment association (GEA) studies are essential to understand the past and ongoing adaptations of organisms to their environment, but those studies are complicated by confounding due to unobserved demographic factors. Although the confounding problem has recently received considerable attention, the proposed approaches do not scale with the high-dimensionality of genomic data. Here, we present a new estimation method for latent factor mixed models (LFMMs) implemented in an upgraded version of the corresponding computer program. We developed a least-squares estimation approach for confounder estimation that provides a unique framework for several categories of genomic data, not restricted to genotypes. The speed of the new algorithm is several order faster than existing GEA approaches and then our previous version of the LFMM program. In addition, the new method outperforms other fast approaches based on principal component or surrogate variable analysis. We illustrate the program use with analyses of the 1000 Genomes Project data set, leading to new findings on adaptation of humans to their environment, and with analyses of DNA methylation profiles providing insights on how tobacco consumption could affect DNA methylation in patients with rheumatoid arthritis. Software availability: Software is available in the R package lfmm at https://bcm-uga.github.io/lfmm/.

Rapid identification and interpretation of gene–environment associations using the new R.SamBada landscape genomics pipeline

“…However, the r software (R Core Team, ) provides an open‐source computing environment adapted to different fields in Biology, in which many of the above‐mentioned pre‐ and postprocessing tasks can be found in various r ‐packages. Further, r can be coupled with compiled languages (such as C++) so as to be more efficient when processing large data sets (see e.g., the case of the software LFMM 2; Caye, Jumentier, Lepeule, & François, , p. 2).…”

In the presence of population structure: From genomics to candidate genes underlying local adaptation

“…To identify SNPs showing significant associations with climate while accounting for population structure, we applied GEMMA (Zhou & Stephens, ) and LFMM 2 (Caye et al, ) on 875 Eurasian accessions (Table ) and four climate variables (Minimum Temperature of Coldest Month; Annual mean temperature; Precipitation during warmest quarter; and Photosynthetically active radiation during fall [time of germination of Italy and Sweden ecotypes (Ågren & Schemske, )]. In addition to identifying SNPs showing significant associations with climate, we also applied BAYESCAN (Foll & Gaggiotti, ) between 40 accessions North Sweden and 25 accessions in South Italy (Tables and ) (referred to as Sweden and Italy hereafter) to identify SNPs that showed significant allele frequency differentiation estimated by F ST after accounting for different models of neutral evolution.…”

“…To identify SNPs showing significant associations with climate, while accounting for population structure we used two prominent methods: GEMMA association (Zhou & Stephens, ) and LFMM (Caye et al, ), version 2. These methods were applied to four climate variables [Minimum Temperature of Coldest Month; Precipitation during Warmest Quarter; Soil moisture; and Photosynthetically Active Radiation during Fall (which is the time of germination for fall ecotypes such as in Italy and Sweden)] important to local adaptation (Lasky et al, ), and a SNP genotype matrix (1001 Genomes Consortium, ) derived from a set of 875 re‐sequenced Arabidopsis thaliana Eurasian accessions (Table ) that excluded laboratory escapees/contaminants (Pisupati et al, ) and accessions from outside the native Eurasian and African range of A. thaliana that may have weaker patterns of local adaptation (Lasky et al, ).…”

Smoothing technical and computational obstacles in gene‐environment associations

“…One of the more urgent extensions to r.sambada to consider would be its integration with other existing gene‐environment associations algorithms already implemented in R, such as lfmm2 (Caye, Jumentier, Lepeule, & François, ). As other methods use different algorithms, they can detect different candidate SNPs, which helps to detect more candidate SNPs (Dalongeville, Benestan, Mouillot, Lobreaux, & Manel, ).…”

“…This would be very helpful for species with no reference genome or a genome that is not fully annotated. I also suggest extending the approach to other types of input genotype data, such as DNA methylation levels to derive associations between epigenetic variation and environment, as recently recommended by Caye et al (). Of course, the candidate SNPs detected in r.sambada still need to be proven to play a role in local adaptation, and this can only be done by designing appropriate experiments (Manel et al, ).…”