Custom 1×1M Agilent microarray designed from E. curvula floral transcriptomes

Description

script.r: R script used for differential expression (DE) analysis of the microarray data. apomictic_vs_sexual.txt: Output file containing the results of the DE analysis. targets.txt: Sample metadata and experimental design description. CUST.fa: FASTA file containing the probe sequences used in the microarray. US23502418_2572215100*: Microarray data files.

Steps to reproduce

Custom Agilent microarray containing around one million probes. The design relied mostly on a floral E. curvula reference transcriptome, but also included EST sequences that had been reported before. Arrays were produced using the SurePrint G3 1x1M format (Agilent). The probes corresponded to 60-mer oligonucleotides designed specifically for this work and, for simplicity, are referred to here as CUST. Probe sequences were obtained from assembled data derived from previous 454 sequencing reads deposited in the NCBI Sequence Read Archive (BioProject 358210). In addition, sequences from the Expressed Sequence Tag (dbEST) database were also included (accession numbers EH183417 to EH195711). The microarray layout and probe design were performed using the Agilent eArray online tool, which allows building customized probe libraries and array formats. For probe design, the Base Composition Methodology was selected, and no linker sequences were used. From the initial set of candidate probes, 970,000 were finally chosen and incorporated into the 1x1M array. The arrays were printed and the slides were supplied directly by Agilent Technologies (Santa Clara, CA, USA). Analysis of microarray data was carried out in R using the limma package from Bioconductor. Raw signal intensities were first corrected for background using the normexp method. Then, quantile normalization was applied between arrays to reduce technical variation, which was noticeable in some cases. A design matrix describing the two experimental conditions was constructed, and linear models were fitted independently for each probe. Differential hybridization between apomictic and sexual samples was evaluated using a contrast that compared both groups directly. To improve variance estimation, standard errors were moderated using an empirical Bayes approach implemented in the eBayes function. P-values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate method. Probes were considered as differentially hybridized when p-values were lower than 0.001 and the absolute log fold change was equal to or greater than 2.

Institutions

• Centro de Recursos Naturales Renovables de la Zona Semiarida

  Buenos Aires, Bahia Blanca

Categories

Licence