Switch branches/tags
Nothing to show
Find file History
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
..
Failed to load latest commit information.
FNN_rate_model_train.m
README.md
wheat.mat

README.md

FNN model for genomic selection in matlab


The implementation of Feedforward Multilayer Neural Network (FNN) model in matlab for genomic selection(GS).

Code download

Depends

  • matlab (>= Version 2015)
  • parpool for parallel (matlab version > 2015)

Contents

Quick start

%you may implement the FNN by the following steps
%clear variables 
clear
FNN_path = './FNN';%set FNN model path
data_path = './wheat.mat';%set dataset path
                          %crossvalind represents for the cross-validation label
                          %pheno represents for the phenotype
                          %SNP represents for the SNP marker matrix
save_path = './FNN_result.mat';%set saved result path
hidden = [35 35];%set hidden neurons
%you must use the MATLAB version no later than MATLAB2015,since the parallel tool used is the parpool
parallel_num = 10;%fit for the 10fold CV, the parallel devices were set to 10
%add the path where the model is located
addpath(genpath(FNN_path))
FNN_rate_model_train(parallel_num,data_path,save_path,hidden)
%you will get a result of FNN_result.mat
%the predict_phenotype represents for the predicted values
%the pheno represents for the real values
%the R represents for the correaltion between the predicted and real values

Ask questions

Please use DeepGS/issues for how to use FNN and reporting bugs