Each T-cell RNA sample from R428 TB patients was co-hybridized with T-cell RNA from a gender-matched and age-matched LTBI to avoid age-dependent or gender-dependent biases. Because of the lower number of NIDs recruited, each NID T-cell RNA was compared with T-cell RNA of an RNA pool from five age-matched and gender-matched TB patients and LTBIs, respectively. Relative comparison of two samples on this type of microarray is based on labelling with different dyes (i.e. Cy3 and Cy5) for co-hybridized samples. To avoid a possible dye-specific bias, an independent swap design [20] was introduced for the hybridizations. Microarrays were processed according to the manufacturer��s instructions, and scanned at 5-��m resolution with an Agilent scanner. Image analysis was performed with feature extraction software (Feature Extractor Version?6.1.1; Agilent Technologies), using default settings and global background normalization. Candidate gene lists were calculated as specified below. RNA was reverse-transcribed to cDNA as described previously [21]. SYBR Green (Applied Biosystems, Foster City, CA, USA) uptake in double-stranded DNA was measured with an ABI PRISM?7900 thermocycler (Applied Biosystems), according to the manufacturer��s instructions. We designed primer pairs with the ABI PRISM primer express Version?2.0.0 software (Applied Biosystems) for real-time qPCR http://www.selleckchem.com/products/jq1.html analysis. The primer pairs for candidate gene verification are listed in Table?2. Glyceraldehyde-3-phosphate E-64 dehydrogenase was used as an internal control for all analyses. Because of limitations in the sample volumes, not all individuals could be included for each candidate gene. Respective numbers are indicated in the figure legends. The Mann�CWhitney U-test or Student��s t-test was used to determine significant differences in qPCR data for candidate genes between study groups. The appropriate test was selected on the basis of Kolmogorov�CSmirnov normality testing (Sigmaplot?10.0.1; Systat, Chicago, IL, USA). Raw microarray data were processed with the function read.maimages of the Bioconductor [22] R package limma [23]. Low-quality spots were detected by the use of eight quality features described in the reference guide of the Agilent Feature Extraction Software (Agilent Technologies, 2008). Then, raw intensities of the spots not flagged out were background-corrected with the norm-exp method [24] of the R package limma. Background-corrected values were Lowess-normalized to obtain as unbiased red/green ratios as possible. For global comparability, the data of all arrays were quantile-normalized [25]. Furthermore, features with mean log intensities <7 were also flagged out, because, for these low-intensity spots, random noise in the data would lead to spurious signals.</div>
↧