Mice cerebellum extracts: SCA7 KI mice vs. wild type mice

Transcriptional profiling of mouse cerebellar extract comparing SCA7 KI mice with wild type mice. Female mice were killed by decapitation on post natal days 10 and 22 and 11 weeks. Goal was to determine gene expression profiles differing between SCA7 KI mice and wild-type mice during post-natal developement of the cerebellum. Gene expression profiling was performed using RNA extracted from the cerebellum of KI and WT mice at P10 (5 WT and 5 KI), P22 (5 WT and 4 KI) and 11 wks (5WT and 6 KI). After labeling, RNAs were hybridized on dual-label G4122F Agilent® chips; a mix of all P10 samples was used as common reference (green channel). Quality control included visual control of the reconstructed image of the chip, M/A plot, corner intensities, outliers, positive and negative intensities, normalization factors. Normalization and statistical analyses were carried out by using BRB-array Tools developed by Dr. Richard Simon and the BRB-ArrayTools development team (Biometrics Research Branch, http://linus.nci.nih.gov/BRB-ArrayTools.html). Genes with less than 50% present calls or with low variability along the arrays (less than 20% of values with at least 2-fold change in either direction from the gene's median value) were excluded from further analysis. For the 1905 remaining probes an interaction between time and genotype was analyzed by regression analysis of the time course of expression. In brief, probes for which variation over time differed for the genotype class were fitted to the following model: log expression ~ time + time**2 + genotype + genotype*time + genotype*time**2. A univiariate p-value < 0.001 (random variance model) was set for significant probes (genotype*time + genotype*time**2) and a False Discovery Rate (FDR) was calculated for each probe (Benjamini & Hochberg, 1995). Differences in profiles were identified with a Self Organisation Tree Algorithm (MultiplExperiment Viewer (MeV), (Saeed et al, 2006). Expression values were averaged by group and then clustered according to their profile as a function of time.本数据集针对小鼠小脑提取物开展转录组分析（Transcriptional profiling），对比脊髓小脑共济失调7型（SCA7）敲入（Knock-In, KI）小鼠与野生型（Wild Type, WT）小鼠的基因表达差异。 实验采用雌性小鼠，分别于出生后第10天（P10）、第22天（P22）以及11周龄时通过断头法处死。本研究的核心目标为解析小脑发育过程中，SCA7 KI小鼠与野生型小鼠之间存在差异的基因表达谱。 针对P10、P22及11周龄三个时间点的KI与WT小鼠小脑提取的RNA开展基因表达谱分析：其中P10组包含5只WT小鼠与5只KI小鼠，P22组包含5只WT小鼠与4只KI小鼠，11周龄组包含5只WT小鼠与6只KI小鼠。RNA标记完成后，将其与双标记G4122F安捷伦（Agilent®）芯片进行杂交；以所有P10样本的混合液作为通用参考样本（对应绿色荧光通道）。 质控流程涵盖芯片重建图像的目视检查、M/A散点图分析、信号角落强度检测、异常值排查、正负信号强度验证以及标准化因子校验。标准化处理与统计分析借助由Richard Simon博士及其开发团队研发的BRB-ArrayTools工具完成（该工具隶属于美国国家癌症研究所生物统计学研究分支，网址：http://linus.nci.nih.gov/BRB-ArrayTools.html）。 对于检出率低于50%，或在芯片中变异程度较低（即相对于基因中位数表达量，至少2倍变化的数值占比不足20%）的基因，将其排除在后续分析之外。针对剩余的1905个探针，通过表达时间进程的回归分析，探究时间与基因型之间的交互效应。简言之，对于随时间变化的表达模式存在基因型差异的探针，采用如下回归模型进行拟合：log(表达量) ~ 时间 + 时间² + 基因型 + 基因型×时间 + 基因型×时间²。针对（基因型×时间 + 基因型×时间²）这一交互项，设定单变量p值＜0.001（采用随机方差模型）作为探针具有显著性的阈值，并针对每个探针计算错误发现率（False Discovery Rate, FDR）（Benjamini与Hochberg，1995）。 利用自组织树算法（借助MultiExperiment Viewer（MeV）工具，Saeed等，2006）识别表达谱差异；首先对各组的表达值进行均值化处理，随后根据其随时间变化的表达模式进行聚类分析。