Computational protocol: Induced Release of a Plant-Defense Volatile ‘Deceptively’ Attracts Insect Vectors to Plants Infected with a Bacterial Pathogen

Similar protocols

Protocol publication

[…] An internal standard of n-octane (300 ng) was added to each sample prior to GC-MS analysis. A one µl aliquot of each extract was injected onto a gas chromatograph (HP 6890) equipped with 30 m×0.25-mm-ID, 0.25 µm film thickness DB-5 capillary column (Quadrex, New Haven, CT, USA), interfaced to a 5973 Mass Selective Detector (Agilent, Palo Alto, CA, USA), in both electron impact (EI) and chemical ionization (CI) modes. Helium was used as the carrier gas in the constant flow mode of 30 cm/sec. The injector was maintained at 260°C. The oven was programmed from 40 to 260°C at 7°C/min. Isobutane was used as the reagent gas for CI, and the ion source temperature was set at 250°C in CI and 220°C in EI. The mass spectra were matched with NIST 2005 version 2.0 standard spectra (NIST, Gaithersburg, MD). The compounds with spectral fit values equal to or greater than 90 and appropriate LRI values were considered positive identifications. When available, mass spectra and retention times were compared to those of authentic standards. Compounds were quantified as equivalents of the total amount of n-octane within each analyzed volatile collection sample. A more accurate quantification of MeSA was made by comparing the total ion chromatograph (TIC) EI response for standard solutions with known quantities of n-octane and MeSA. A response factor of 1.2 was then established for MeSA relative to the n-octane based values.The resulting volatile profiles were standardized as equivalents of n-octane within each sample analyzed. The characteristic set of variables that defined a particular group (e.g. non-infected versus infected plant) was found using the ‘varSelRFBoot’ function of the package ‘varSelRF’ for the ‘randomForest’ analysis (R software version 2.9.0, R Development Core Team 2009). We used the varSelRF algorithm with Random Forests to select the minimum set of VOCs that were characteristic of differences between infected and non-infected plants. The tree-based Random Forests algorithm performs hierarchical clustering via multi-scale and combinatorial bootstrap resampling and is most appropriate for data where the variables (VOCs in this case) outnumber the samples, and where the variables are auto correlated, which is a typical problem of conventional multivariate analysis of such data. Consequently, this type of analysis is common in bioinformatics, chemoinformatics and similar data-rich fields. We employed 200 bootstrapping iterations of the Random Forests algorithm to arrive at a minimal set of VOCs that could differentiate between infected and non-infected plants. We also calculated the mean decrease in accuracy (MDA) when individual VOCs are removed from the analysis. MDA values indicate the importance value of particular VOCs for the discrimination between treatments. […]

Pipeline specifications

Software tools varSelRF, randomforest
Applications Drug design, Miscellaneous
Organisms Candidatus Liberibacter asiaticus, Diaphorina citri
Chemicals Boron, Iron, Nitrogen, Phosphorus, Potassium, Sulfur, Zinc