Computational protocol: Single-Species Microarrays and Comparative Transcriptomics

Similar protocols

Protocol publication

[…] We performed new expression analyses on testis and brain tissue from XL, XB, and HXLXB. For each tissue from each species or hybrid, RNA was isolated using TRIzol® Reagent (Invitrogen Life Technologies) according to the manufacturer's protocol, purified with RNeasy Mini Kit (Qiagen), and its integrity assessed on an Agilent BioAnalyzer. Two micrograms of total RNA was used to prepare biotin labeled cRNA probes, which were subsequently hybridized to Affymetrix Xenopus laevis expression arrays following the manufacturer's protocol.We performed new gDNA hybridizations using gDNA from XL, XB, and XM and compared these to gDNA hybridizations on XL and XM that were performed by Malone et al. . For our gDNA hybridizations, five micrograms of gDNA from each species was fragmented with Dpn I at 37°C for 3 hours. Fragmented gDNA was purified with Qiagen PCR clean-up kit and the fragment distribution was checked on Agilent Bioanalyzer (Agilent) using the DNA 1000 assay. 50–100 nanograms of fragmented gDNA were then amplified using the BioPrime Labeling System (Invitrogen) following the manufacturers instructions. After completion of the Klenow Pol I catalyzed reaction, the distribution of PCR products was examined on Agilent Bioanalyzer with the DNA 1000 kit. The entire volume of the product (∼50 μl) was used in the hybridization reaction on the Affymetrix Xenopus laevis Gene Chip. Hybridization, staining, washing and scanning were performed as described in the Expression Analysis Technical Manual. This protocol is similar to that used by Hammond et al. .After scanning, raw expression data were converted into CEL files using Microarray Analysis Suite version 5 (MAS 5, Affymetrix). For each pairwise comparison, CEL files were pre-normalized with the Robust Multichip Average (RMA) algorithm in RMAexpress using custom CDF files (probemasks) and the default parameters, which include a median polish and quantile normalization. The normalized data were used in the R statistical package following the protocol in . An empirical Bayesian model was used to compute a moderated t-statistic using the limma package from Bioconductor . The TopTable function gave a P-value for differential expression for each gene that was adjusted using the Benjamini and Hochberg method to control for the false discovery rate. cDNA and gDNA hybridizations that we performed have been deposited in the Gene expression omnibus database , GEO Series accession number GSE12625. We also analyzed other data from this database (GSM241082-4 , GSM99995-7 , GSM99980-2 ). Expression data and genomic hybridizations from XL and XM testis and ovary that were not found in GEO were kindly provided by Pawel Michalak.We used a re-sampling approach to test whether the proportion of divergently expressed genes in different analyses (each with a unique number of genes analyzed) were significantly different. Given two analyses with w and x genes of which y and z are significantly divergently expressed, respectively, using a PERL script we generated 1000 simulated datasets, each with w genes, by re-sampling a distribution of (w+x) total genes with (y+z) genes that are significantly divergently expressed. Where (y/w)<(z/x), the two-sided probability of the null hypothesis of no difference is twice the proportion of these simulated datasets that had a proportion of divergently expressed genes lower than y/w (i.e. more different from z/x). Because some of the genes in these different analyses are the same and should therefore have correlated expression levels, the inclusion of these genes in this comparison reduces the power to reject the null hypothesis, making this test conservative. […]

Pipeline specifications

Software tools RMAExpress, limma
Databases GEO
Application Gene expression microarray analysis
Organisms Xenopus laevis, Xenopus borealis