View source for IC4R004-Metabolomics-2015-25766578

== Project Title ==
'''A targeted metabolomics approach toward understanding metabolic variations in rice under pesticide stress'''

==The Background of This Project==
[[File:IC4R004-Metabolomics-2015-25766578-1.png|right|thumb|227px|'''Figure 1. O,O-Diethyl (IUPAC). O-2-isopropyl-6-methyl-pyrimidin-4-yl phosphorothioate''']]
* Rice is considered as the most important food for populations of developing countries and is the main dish to half of the world’s population [1]. It is an annual grass of the Gramineae family belonging to the Oryza genus [2]. Oryza sativa L. is grown all over the world and is cultivated in humid and temperate environments, making its production susceptible to fungi, insects, and mites. More than 70 insect species have been recorded as rice pests, so this would be considered as one of the major constraints on cropyields causing serious reduction in plant production or even blockage of it. To address this problem, several kinds of insecticides, fungicides, and herbicides are used in order to protect crops against pest damage. The application of pesticides in rice fields has become a popular approach to controlling pest damage during early periods of rice cultivation in Asia [3]. Diazinon (O,O-diethyl O-2-isopropyl-6-methylpyrimidin-4-yl thiophosphate; Fig. 1) is an organophosphorus pesticide that was commercially introduced in 1952 [4]. Thanks to its inhibitory effects on acetyl cholinesterase enzyme in a vast majority of insects, diazinon is universally used in agricultural sectors for plant protection against a variety of sucking and leaf-eating insects [5].
* Nevertheless, several reports have demonstrated that diazinon is immunotoxic [6], cytotoxic [7], and genotoxic [8]; hence, it exhibits toxic properties and potential risk to human health. Because rice is one of the most popular crops worldwide, it seems essential to probe the influence of diazinon on metabolite profiling of rice. To this end, metabolomics is one of the most powerful tools for providing an overview of metabolite changes under various abiotic stresses, namely pesticide stress [9]. So far, no report regarding the influence of diazinon on metabolite profiling of rice has been recorded. 
[[File:IC4R004-Metabolomics-2015-25766578-2.png|left|thumb|527px|'''Figure 2. The OOB error rate shows the prediction accuracy of classification.''']]
* '''In this project, the researchers investigate the effect of diazinon on rice metabolite profiling under subtropical climatic conditions. The aim of the current study was to classify and follow the trea- ted group and the control group, which would detect possible main variations in metabolite profiling of rice under diazinon stress, based on the data acquired from mass spectroscopy as well as retention times of the peaks obtained from MS workstation software.'''

==Plant Materials==
* Seeds of Shiroodi variety (O. sativa ssp. indica) were obtained from the Rice Research Institute of Iran and were subsequently cul- tivated in the same open field to avoid the influence of growing location. Diazinon was employed to prevent the plant from insect damage, whereas its applied concentration was based on the recommended permitted dosage from the Plant Protection Organization. Plants were subjected to 10% granular formulation of diazinon during heading and flowering time. Untreated plants were planted under the same experimental con- ditions. Rice leaves were taken from control and treated plants at 24, 48, 72, 96, and 120 h after treatment. Seven replicates at each time point were collected from different plants and immediately chilled in liquid nitrogen. Frozen leaves were manually ground in a mortar using liquid nitrogen in order to keep samples at cryogenic temperature, and eventually all samples were stored at -80 °C until metabolite analysis.
==Research Findings ==
[[File:IC4R004-Metabolomics-2015-25766578-3.png|right|thumb|427px|'''Figure 4. Variable importance for RF classification of control and exposed rice plants.''']]
* Due to the observation of various intensities in GC–MS data, especially for following metabolites with a wide range of concen- trations, it would be necessary to calibrate and normalize the peaks. Thus, GC–MS data set related to the treatment of rice with diazinon was split into training and test sets using random division. To well illustrate the performance of the proposed approach for two metabolomics datasets (treated and control), our models should be validated by predicting the classes of the test set that was not used in training set. The model based on random selection was constructed on the training set, which contained approximately two-thirds of the samples (7 replicates × 5 time points = 35 samples). Each GC–MS spectrum contained 135 features that represented different retention times in GC chromatograms of rice metabolites in treated and control samples. These features were considered as descriptive variables in the classification model, whereas the class numbers of the different samples were employed as a response.
* In creating each tree in RF, different bootstrap samples from the original data are applied. In the construction of the k-th tree, approximately one-third of the cases are not used and are left out of the bootstrap sample. These samples as out-of-bag (OOB) turn out to be useful. In the construction of the k-th tree, with the purpose of developing a classification, each OOB example put down the k-th tree. An estimate of the test set error is obtained by comparing this classification with the class label present in the data. Then, it is often claimed that the OOB is an unbiased esti- mate of the true prediction error [17]. The OOB data were used to estimate the prediction accuracy of GC–MS data classification. Fig. 2 presents the OOB error rate.
* To assess the importance of each variable, the RF model can provide the measure of variable importance. To measure the importance of the m-th variable in the left out cases for the k-th tree, all values of the m-th variable transpose randomly. In the next step, these new variable values put down the tree and get classifi- cations. By minus the proportion of votes for its true class minus the maximum of the proportion of votes for each of the other classes, its margin at the end of a run is calculated for the i-th case in the data. In this way, the importance of the m-th variable is mea- sured by the average lowering of the margin across all cases when the m-th variable is randomly permuted. '''Figure 4''' illustrates the contribution of each variable to the cluster analysis model. In this study, RF identifies the most prominent variables between control and treated rice under diazinon stress.

==Labs working on this Project==
* Department of Phytochemistry, Medicinal Plants and Drugs Research Institute, Shahid Beheshti University, Tehran, Iran
* Department of Chemistry, Sharif University of Technology, Tehran, Iran

==Corresponding Author==
'''Alireza Ghassempour''' (E-mail: a-ghassempour@sbu.ac.ir)