MDC/CAS/ICL RAE230A Kidney Database RMA 2ZPlus8 (April/05 freeze)

Accession number: GN65

    Summary:

This April 2005 data set provides estimates of mRNA expression in normal kidneys of 32 strains of rats. The set includes the hypertensive SHR strain, the normotensive BN strain, and 30 HXB/BXH recombinant inbred strains. Each strain was sampled in quadruplicate (6-week-old males). Animals and tissues were generated by Michal Pravenec and colleagues at the Czech Academy of Sciences (CAS). RNA samples were processed at the Max-Delbrück-Center (MDC), Berlin Buch by Nobert Hübner and colleagues. Transcriptome mapping was carried out by Timothy Aitman and colleagues at the Imperial College, London (ICL). Samples were hybridized individually to a total of 128 Affymetrix RAE230A array. This particular data set includes 120 arrays processed using the RMA protocol. RMA values of each array were adjusted to an average of 8 units and a standard deviation of 2 units (2ZPlus8). This data set complements the MAS5 data set exploited by Hübner and colleagues 2005. Download the particular transform in an Excel work book with both strain means and SEMs.

These data can also be viewed using the eQTL Explorer Java application by John Mangion, Tim Aitman, and colleagues (Mueller et al. 2006).

    About the cases used to generate this set of data:

We have exploited a set of HXB/BXH recombinant inbred strains of rats generated over the past 20 years in Prague. The parental strains from which all HXB lines are derived are SHR (SHR/OlaIpcv or HSR = H) and Brown Norway (BN.Lx/Cub= B). These parental strains have been used extensively to study cardiovascular system physiology and genetics.

The HXB strains were generated by Michal Pravenec at the Institute of Physiology, Czech Academy of Sciences. The BXH strains were generated by Vladimir Kren (see Pravenec et al. 1989, 2004) at a similar animal facility at the Institute of Biology and Medical Genetics, 1st Medical Faculty, Charles University. These strains are at approximately the 6oth geenration of continuous inbreeding (F60).

Animals used in the transcriptome analyses of kidney and fat (Hübner and colleagues, 2005) were weaned at 4 weeks. Those born at the Charles University were transferred to the Institute of Physiology. Animals were reared on a commerical rat chow (ST-1 from VELAZ, Czech Republic). Four males were house per cage. Cages were made of polystyrene and have a floor size of 22 x 38 cm and height of 23 cm. The bedding was changed twice a week. Light cycle was 12:12 on-off. Vivarium rooms were maintained at 23 deg. C. Rats were sexually naive. All males used in the initial transcriptome studies (Hubner et al., 2005) were born between May and August 2002. They were sacrificed unfastged by rapid cervical dislocation between 9 and 10 AM, following an approved animal protocol (Ethics Committee of the Institute of Physiology, Czech Academy of Sciences, Prague; Animal Protectiion Law of the Czech Republic (311/1997).

    About the tissue used to generate these data:

All tissues were collected at the age of 6 weeks. Kidneys and other organs were rapidly dissected and cleaned of fat, inserted into a vial, and immersed in liquid nitrogen for storage until RNA extraction.

The table below lists the arrays by strain and sample identifier. Each array was hybridized with mRNA from a single young male rat.

*: These eight arrays were excluded in the final strain summary data. See section of Quality Control for further explanation.

    About the array platform:

Affymetrix 230A GeneChip: Expression data were generated using 230A array. The chromosomal locations of probe sets were determined by BLAT analysis of concatenated probe sequences using the Rat Genome Sequencing Consortium assembly.

    About Quality Control Procedures:

RNA processing:RNA was extracted using Trizol reagent (Invitrogen) and purified using an RNeasy Mini kit from Qiagen. Double-stranded cDNA was generated without pooling. The Ambion MEGAscript T7 kit from Ambion was used to generate biotinylated cRNA for kidney. Fat samples were processed at this step using the Enzo Diagnostics Bioarray High Yield RNA Transcript labeling kit. See Hubner et al. 2005 for additional detail. One-hundred and twenty eight samples passed RNA quality control assays.

Probe level QC: All 128 CEL files were collected into a single DataDesk 6.2 analysis file. Probe data from pairs of arrays were plotted and compared. Eight arrays were considered potential outliers (despite having passed RNA quality control) and in the interest of minimizing technical variance, a decision was made to withhold them from the calculation of strain means used in WebQTL. The remaining 120 arrays were quantile normalized and reexamined in DataDesk to ensure reasonble colinearity of all array data sets.

Probe set level QC: Probe set level QC involves counting the number of times that a single array data set from a single sample generates outliers at the level of the probe set consensus estimate of expression. With 120 arrays, any single array should generate a comparatively small fraction of the total number of outlier calls. This final step of array QC has NOT been implemented in this data set.

    About data processing:

Probe set data: The original CEL files were read into the R environment (Ihaka and Gentleman 1996). Data were processed using the Robust Multichip Average (RMA) method (Irrizary et al. 2003). Values were log2 transformed. Probe set values listed in WebQTL pages are typically the averages of four biological replicates within strain.

This data set include further normalization to produce final estimates of expression that can be compared directly to the other transforms (average of 8 units and stabilized variance of 2 units within each array). Data were analyzed as follows:

• Step 1: RMA values were generated as described above. These values already incorporate the quantile normalization
• Step 2: We computed the Z scores for each value.
• Step 3: We multiplied all Z scores by 2.
• Step 4: We added 8 to the value of all Z scores. The consequence of this simple set of transformations is to produce a set of Z scores that have a mean of 8, a variance of 4, and a standard deviation of 2. The advantage of this modified Z score is that a two-fold difference in expression level corresponds approximately to a 1 unit difference.
• Step 5: Finally, we computed the arithmetic mean of the values for the set of microarrays for each strain. We have not corrected for background beyond the background correction implemented by Affymetrix.

Please see Bolstad and colleagues (2003) for a helpful comparison of RMA and other common methods of processing Affymetrix array data sets.

    Data source acknowledgment:

This work was supported with funds to TJA by the MRC Clinical Sciences Centre, the British Heart Foundation, and the Wellcome Trust Cardiovascular Functional Genomics Intiative; to NH from the German Ministry for Science and Education (National Genome Research Network); to MP and vladimir Kren from the Grant Agency of the Czech Republic; to MP and TJA from the Wellcome Trust Collaborative Research Initiative grant, to Theodore W Kurtz from the NIH, to TWK and MP from a Fogarty International Research Collaboration Award. Microarrays were a generous donation of Affymetrix Inc.

    Information about this text file:

This text file originally generated by Robert Williams, Norbert Hübner, Michal Pravnec, Timothy Aitman, April 19, 2005. Updated by RWW, April 20, 2005; April 28, 2005.