NirsPredict 1.0


NirsPredict is an application allowing to make predictions about phenotypic traits of Arabidopsis thaliana by submiting related NIRS spectra using deep-learning. NirsPredict has three functionalities :

  • The submition of your data to our deep-learning algorithms to get traits predictions.
  • The consultation of the database used to construct our deep-learning models, it's possible to target the data you're interested in with some filters and download the corresponding dataset , morevover graphical representation of the dataset will be printed.
  • Contribute to the application by submiting your dataset so we can examine it and add it to our database/models

NirsPredict Application

Download application manual
Download metadata guide

Predict phenotypic traits

Submitted files must have headers

A dataset with at least 100 spectra is advised to get high-quality predictions

Download and consult the application manual above to get more details on input format and functionnalities


The prediction's robustness and the number of input to provide will rely on the selected mode.


Select some traits or/and metabolites to predict. Beware that the more you select the longer it will be.

You must provide a valid email adress before launch so you can receive your run's results
Prediction with our models can be imprecise due to specific conditions of your samples. In this case you can train against our deep learning scripts to build new model more suitable for your data to make predictions.

Visualize and download data from the database built from Vasseur et al. (2022)

General Informations

to

Sample Informations

Output

Choose query output format


                            
                            
                              
                              Download
                            
                          

                        

                        

                        

                          
Additional traits

                          
Customize the features present in the provided dataset

                          

                            

                              

                                

                                  
                                  
                                

                              

                              

                                

                                  
                                  
                                

                              

                            

                            

                              

                                

                                  
                                  
                                

                              

                              

                                

                                  
                                  
                                

                              

                            

                          

                        

                        

                        
Complete predictions dataset

                        

                          

                            

                              
We have used the application to predict values for every traits and from every spectra contained in the database.


                                                                                                       
All known values have been replaced by predicted values to make a complete predicted dataset.

                              
                                
                                Download complete predictions file
                              
                            

                          

                        

                      

                      

                        

                          

                            
                            

                              
                            

                            

                              
                            

                          

                        

                      

                    

                    

                      

                        

                        
Share your data

                        

                          
The submitted dataset will be examined and if relevant it may be integrated to the database to extend it and make future predictions more precise

                          

                            
                            

                              
                              
                            

                            

                              

                            

                          

                          

                            

                              
                              

                              You must provide a valid email adress before send so we can ask you further informations if your data is relevant
                            

                            

                              

                                
                                
                              

                              
                            

                          

                        

                      

                    

                  

                

              

            

          

          

            

              

                

                  
NirsPredict

                  
NirsPredict is a tool allowing to make deep-learning predictions about phenotypic traits of A. thaliana by submiting related NIRS spectra

                  
'A Perspective on Plant Phenomics: Coupling Deep Learning and Near-Infrared Spectroscopy' - Vasseur et al.

                  
You can contact us on this email adress : NirsPredict@post.com 

                  
How to cite: Comming soon