DeepGenomeScan : A Deep Learning Approach for Whole Genome Scan (WGS) and Genome-Wide Association Studies (GWAS)

This package implements the genome scan and genome-wide association studies using deep neural networks (i.e, Multi-Layer Perceptron (MLP), Convolutional Neural Network (CNN)). DeepGenomeScan offers heuristic computational framework integrating different neural network architectures (i.e.,Multi-Layer Perceptron (MLP), convolutional neural network(CNN)) and robust resampling methods, as well as the Model-Agnostic interpretation of feature importance for convolutional neural networks. DeepGenomeScan, in other words, deep learning for genome-wide scanning, is a deep learning approach for detecting signatures of natural selection or for performing omics-based genome-wide association studies, such as GWAS, PWAS, TWAS, MWAS. The framework makes the implementation user-friendly. It is compatible with most of the self-defined machine learning models (the self-defined models should be complete, including tunable parameters, fitted model, predicted model, examples can be found in our tutorial). Users can adopt this framework to study various evolutionary questions.

Install packages

Dependencies and environment requirements

Note: Environment requirements: python should be installed and the python package of Keras and Tensorflow should also be installed and work properly with the system

library("devtools")

## The modified version of caret
devtools::install_github("xinghuq/CaretPlus/pkg/caret")

###DA and KLFDAPC are used for SpGenomeScan


 if (!requireNamespace("DA", quietly=TRUE))
 
  devtools::install_github("xinghuq/DA")

requireNamespace("KLFDAPC")

 if (!requireNamespace("KLFDAPC", quietly=TRUE))

  devtools::Install_github("xinghuq/KLFDAPC")
  

if (!requireNamespace("keras", quietly=TRUE))

 Install.packages("keras")
  
 if (!requireNamespace("tensorflow", quietly=TRUE))
 
  install.packages("tensorflow")
 if (!requireNamespace("kerasR", quietly=TRUE))
 
  install.packages("kerasR")

### The latest version of DeepGenomeScan 

devtools::install_github("xinghuq/DeepGenomeScan")

If using Tensorflow or keras model (see example in tutorial), checking the python environment

library("rappdirs")
library("reticulate")
reticulate::use_python("/usr/bin/python3")
library(caret) ### for ML calling functions and performance estimation, users should use the modified version at xinghuq/CaretPlus/caret instead of the original version
library(keras)  
library("tensorflow")

# checking if Tensorflow works properly
K0=keras::backend()

loading library

library("DeepGenomeScan")
library("caret")### for ML calling functions and performance estimation
library("keras") ### for DL
library("tensorflow")
library("caretEnsemble")
library("kerasR")
library("RSNNS")
library("NeuralNetTools")

Example

Preparing data

f <- system.file('extdata',package='DeepGenomeScan')
infile <- file.path(f, "sim1.csv")
sim_example=read.csv(infile)
genotype=sim_example[,-c(1:14)]
env=sim_example[,2:11]
str(sim_example)

Setting the resampling method

econtrol1 <- trainControl(## 5-fold CV, repeat 5 times
  method = "adaptive_cv",
  number = 5,
  ## repeated ten times
  repeats = 5,search = "random")
set.seed(999)
options(warn=-1

DeepGenomeScan with “mlp” model

GSmlp<- DeepGenomeScan(as.matrix(genotype_norm),env$envir1,
                                  method="mlp",
                                  metric = "RMSE",## "Accuracy", "RMSE","Rsquared","MAE"
                                  tuneLength = 10, ### random search 10 hyperparameter combinations
                                  # tuneGrid=CNNGrid, or setting the tune grid
                                  trControl = econtrol1)

#### varIMP for SNPs
out=varImp(GSmlp,scale = FALSE)

Plot the SNP importance scores

#### Plot only the importance, remember to scan all environmnet factors and use DLqvalue function to convert the multi-effect importance values to q-values
### here is only plot importance of an example of using one environment factor
 scaled_imp=normalizeData(out$importance$Overall,type = "0_1")
SNPimp<-data.frame(index = c(1:1000), MLP= -log10(scaled_imp))
plot(y=-SNPimp$MLP,x=1:1000L,  ylab="SNP importance")
abline(h=2, col="blue")

Welcome any feedback and pull request.

Version

The current version is 0.5.5 (Sep 2, 2020).

Citation

Qin, X. 2020. DeepGenomeScan: A Deep Learning Approach for Whole Genome Scan (WGS) and Genome-Wide Association Studies (GWAS). R package, v0.5.5.

Qin X, Chiang CWK, Gaggiotti OE. 2021. Deciphering signatures of natural selection via deep learning. bioRxiv:2021.2005.2027.445973.