High-throughput informatics for antibodies

描述(由申请人提供)：免疫球蛋白(Ig)基因座的体细胞超突变(SHM)是产生抗体多样性的基本过程。用于分析免疫球蛋白基因突变谱的高通量方法，如罗氏454深度测序，仍然不准确，部分原因是后处理需要专门的生物信息学。最近对Illumina MiSeq平台的改进允许成对端2x300个核苷酸读取，这将以比454平台低60倍的成本实现更准确的IGV测序。随着成本的进一步下降，像MiSeq这样的平台将成为常见的实验室设备，但只有在适当的生物信息学工具可用的情况下才会有用。准确的免疫球蛋白基因深度测序正迅速成为广泛临床应用的标准，包括确定预后和检测B细胞恶性肿瘤的微小残留病，自身免疫性疾病的特征和评估疫苗反应。在目标1中，我们将开发一个用户友好的生物信息学管道(SHMPrep)，以改进对来自Illumina MiSeq平台的IGV序列的突变要求。我们假设，对独立的PCR和测序误差影响进行统计建模可以将MiSeq IGV序列的质量提高到与Sanger测序相当的水平。该管道将与以前开发的分析工具(SHMTool)集成，以允许非计算机专家，如大多数临床医生，在普通台式计算机上处理MiSeq IGV数据集。随着高吞吐量数据的积累，对我们已经拥有的数据拥有分析方法变得更加重要，而不是产生更多的数据。免疫球蛋白基因突变谱取决于多种因素，包括碱基组成、激活诱导脱氨酶热点和冷点的丰度和位置、POL-η热点组成和总突变频率。这种复杂性使得比较不同免疫球蛋白区域的突变谱变得困难。在目标2中，我们将开发统计方法来比较不同的IGV区域，考虑到序列组成以及突变饱和度和链偏差，这对于识别免疫缺陷(如艾滋病)、B细胞恶性肿瘤和其他癌症的修复缺陷是重要的。我们仍然对IGHV基因之间的差异知之甚少。为什么有如此多的V区，以及特定的免疫球蛋白基因和免疫反应之间如此强烈的关联？在目标3中，我们将开发一个统计模型来预测突变频率，该模型将允许表示已知的分子相互作用，例如，AID靶向和容易出错的错配修复之间的相互作用。SHMTool将使用模型中预测的突变频率，在没有对照数据集的情况下提供一个比较基准。该模型将被用来在比以前更深的水平上描述每个IGHV基因的特征，从而允许跨物种比较。从长远来看，这样的模型将有助于更好地理解IGHV基因和谱系的进化变化。

DESCRIPTION (provided by applicant): Somatic hypermutation (SHM) of the Immunoglobulin (Ig) loci is a fundamental process in generating antibody diversity. High-throughput methods for profiling mutation spectra of Ig genes such as Roche 454 deep sequencing remain inaccurate and are expensive in part due to specialized bioinformatics required for post-processing. Recent improvements to the Illumina MiSeq platform allowing paired-end 2x300 nt reads will enable more accurate IgV sequencing at a >60 times lower cost than the 454 platform. As costs fall further, platforms such as MiSeq will become common lab equipment, but will only be useful if the appropriate bioinformatics tools are available. Accurate deep sequencing of the Ig loci is rapidly becoming the standard in a broad range of clinical applications including determining prognosis and detection of Minimal Residual Disease in B cell malignancies, characterization of autoimmune diseases and evaluating vaccine responses. In Aim 1 we will develop a user-friendly bioinformatics pipeline (SHMPrep) to improve mutation calls for IgV sequences from the Illumina MiSeq platform. We hypothesize that statistical modeling of independent PCR vs sequencing error effects can improve the quality of MiSeq IgV sequences to levels comparable to Sanger sequencing. The pipeline will be integrated with a previously developed analysis tool (SHMTool) to allow non computer experts, such as most clinicians, to process MiSeq IgV datasets on an ordinary desktop computer. As high-throughput data accumulates it becomes more important to have analysis methods for the data we already have rather than producing yet more data. IgV mutation spectra depend on many factors including base composition, abundance and location of activation induced deaminase (AID) hot and cold spots, Pol-η hot spot composition and overall mutation frequency. This complexity makes it difficult to compare mutation spectra from different IgV regions. In Aim 2 we will develop statistical methods for comparing different IgV regions taking into account sequence composition as well as mutation saturation and strand bias, which is important in identifying repair defects in immunodeficiencies such as AIDS and in B-cell malignancies and other cancers. We still understand little about the differences between the IGHV genes. Why are there so many V regions and such strong associations between particular Ig genes and immune responses? In Aim 3 we will develop a statistical model for predicting mutation frequencies that will allow known molecular interactions to be represented, for example, the interaction between AID targeting and error-prone mismatch repair. Predicted mutation frequencies from the model will be used by SHMTool to provide a comparative benchmark in situations where no control dataset is available. The model will be used to characterize each IGHV gene at a deeper level than was previously possible, allowing cross-species comparisons. In the longer term such a model will facilitate a better understanding of evolutionary changes in the IGHV genes and repertoire.