Salinity induced gene expression profiling in the rice genotypes using Agilent Rice Gene Expression 4x44K Microarray

Stress was imposed on 14 day old seedlings and total RNA from the whole seedlings was collected after 48 h of stressed period (i.e., from 16 day old seedlings). The transcriptomic data of these genotypes (submitted now) along with the previously submitted dataset (GSE79043) were used to develop traditional and orthogonal signal correction filtered partial least square discriminant analysis (OSC-PLSDA) models using the differential transcriptomic fingerprints in tolerant vs susceptible rice genotypes.

These models are then used to test the transcriptomic fingerprints of stressed and unstressed samples of several genetically diverse rice genotypes in a way to predict their salinity tolerance status. The accuracy of the prediction using OSC-PLSDA model was higher compared to that of the traditional PLSDA model. Besides, the accuracy of the prediction of salinity tolerance of the salt stressed samples was higher than that of the unstressed samples indicating that the stress induced transcriptomic fingerprints can better discriminate the tolerance and susceptible genotypes.The minimum number of genes (109 genes) that highly contributed to the classification and model prediction of salinity tolerance status was identified using the Variable Importance vector in OSC-PLSDA model.This study provides an example of prediction model building using whole genome transcriptomic data.