Developing New Statisical Methods to Detect Common and Rare Variants Involved in Neuropsychiatric Disorders

我们的目标是开发新的分析工具，可以比现有的方法更有效、更准确地识别遗传变异，然后在精神疾病或其他复杂的人类疾病的数据集中进行测试。我们一直在与校内和校外的合作者积极合作。一些例子包括：1)利用基于途径的方法检测影响强迫症（OCD）患者血清素再摄取抑制剂药物治疗选择的遗传结构域；2)在拉丁裔人群（精神分裂症和躁郁症）中建立有效的外显子组测序分析管道；3)开发一种有效的检验两个连续变量的非线性相关性的方法。这为检验两个连续变量之间的非线性相关性提供了一种新的方法。有必要改进软件的当前版本，以便使用基于linux的计算机或台式机的个人可以使用它来运行他们的分析。

Our goal is to develop new analytic tools that can be used to identify genetic variation more efficiently and accurately than the existing methods and then test them in the datasets of psychiatric disorders or other complex human diseases. We have been actively working with both intramural and extramural collaborators. A few examples include 1) Using pathway-bases approaches to detect genetic domains underlying the pharmacotherapy options of serotonin reuptake inhibitors effect in study subjects affected with obsessive compulsive disorder (OCD); 2) Establishing efficient pipelines for exome sequencing analyses in a Latino population (schizophrenia and bipolar); and 3) Developing an efficient test for nonlinear dependence of two continuous variables. This presents a new way of testing nonlinear dependence between two continuous variables. There is a need to improve the current version of the software so that it can be used by individuals who use Linux-based computer or desktops to run their analysis. Summary We collaborated with extramural scientists and developed a new statistical method called CANOVA. Using this method, one can generalize the within category variance in traditional analysis of variance (ANOVA). Using extensive simulations, we extensively evaluated the performance of CANOVA. We then applied CANOVA to a real dataset and showed that the power of CANOVA performs better when the correlation is highly non-linear. In addition, we developed a novel mixture model to estimate the time to antidepressant effect onset and its association with covariates such as age, gender and baseline anxiety. We evaluated the model's overall utility and performance via extensive simulations. We demonstrated its use by application to a longitudinal dataset from the Sequenced treatment Alternatives to Relieve Depression (STAR*D) study. Our algorithm successfully identified age and anxiety status as significant factors in influencing the onset distribution of citalopram. And we developed a pathway-based pipeline which can be used to link genome-wide association study signals to several important biological pathways. We used our own pipelines to analyze drug response on Obsessive Compulsive Disorder (OCD) GWAS data to select significant pathways. In a recent publication, we reported the suggestive roles of genes in the glutamatergic neurotransmission system and the serotonergic system. The results presented may provide new insights into genetic mechanisms underlying treatment response in OCD. More recently, we applied the same approach to seeking genetic variants underlying antidepressants such as ketamine. The potential findings will be reported in the future.