NIRSpredict: a platform for predicting plant traits from near infra-red spectroscopy

Abstract

Near-infrared spectroscopy (NIRS) has become a popular tool for investigating phenotypic variability in plants. We developed the Shiny NIRSpredict application to get predictions of 81 Arabidopsis thaliana phenotypic traits, including classical functional traits as well as a large variety of commonly measured chemical compounds, based from near-infrared spectroscopy values based on deep learning. It is freely accessible at the following URL: https://shiny.cefe.cnrs.fr/NirsPredict/ . NIRSpredict has three main functionalities. First, it allows users to submit their spectrum values to get the predictions of plant traits from models built with the hosted A. thaliana database. Second, users have access to the database of traits used for model calibration. Data can be filtered and extracted on user's choice and visualized in a global context. Third, a user can submit his own dataset to extend the database and get part of the application development. NIRSpredict provides an easy-to-use and efficient method for trait prediction and an access to a large dataset of A. thaliana trait values. In addition to covering many of functional traits it also allows to predict a large variety of commonly measured chemical compounds. As a reliable way of characterizing plant populations across geographical ranges, NIRSpredict can facilitate the adoption of phenomics in functional and evolutionary ecology.

Keywords: Arabidopsis thaliana; Functional traits; Genetic variability; Machine learning; Phenomics; Secondary metabolites; Trait prediction.

Fig. 1

Diagram of the architecture of the neural network used by NIRSpredict. Diagram of the architecture of the convolutional neural network used to calibrate Arabidopsis thaliana near-infrared spectra prediction models. It was generated using Netron. Roeder, L. (2023, November 15). lutzroeder/netron GitHub repository. Retrieved from https://github.com/lutzroeder/netron

Fig. 2

Representative diagramm of NIRSpredict features. Schema of the application showing how the application works. The available running options of the prediction features are represented with their needed input and their expected output. The database query process is represented with the available data format and the associated charts

Fig. 3

Screenshots showing the NIRSpredict main pages in use. (a) Predictions tab with uploaded files and fields fulfilled ; (b) Result of a query on the database matching water stress/high temperature treatment condition with graphics outputs. Mean absorption comparison between the subset and the database values. Principal component analysis of the subset spectrum values compared to the database values one

Fig. 4

Graphical analysis associated to the predictions. Linear regression of predictions values. Predicted values have been obtained following the creation of a model and the use of a calibration and a validation dataset. The predicted values are compared to the observed values for the leaf thickness trait. The values follow the x = y pattern thus showing a correct prediction accuracy