STRUCTURAL VISUALIZATION IN BIOINFORMATICS

本子项目是利用由NIH/NCRR资助的中心赠款提供的资源的众多研究子项目之一。子项目和研究者（PI）可能已经从另一个NIH来源获得了主要资金，因此可以在其他CRISP条目中表示。列出的机构是中心的，不一定是研究者的机构。这笔赠款的具体目的如原提案所述。由于这项资助的期限是3年而不是5年，因此研究人员将更多地专注于具体分析和可视化技术的开发，而不是像最初提议的那样专注于构建通用数据处理基础设施。赠款的第一年将专门用于获取基本知识和实施基本软件基础设施。今后的工作将按照建议的概述继续进行。人员：赠款的资金直到2004年11月才正式到位，将最初的工作人员招聘推迟到2005年春季。然而，在2004年秋季学期，PI与两名少数民族学生合作，专门针对以下具体目标的两个方面进行了项目。为了弥补由于这些延误而损失的时间，研究者在2005年1月雇佣了两名兼职研究生研究助理（Egle Pilipaviciute和Dragana Veljkovic）和两名本科程序员（Jason Edwards和James Packer）。这个全职研究软件开发人员的职位于2005年3月初由科里·伯克哈特（Cory Burkhardt）担任，他是UTSA的优等毕业生。调查员还雇了大卫·比格姆，刚从德州大学毕业即将进入计算机科学硕士？2005年夏天，他在美国的一个项目中工作了三个月。David将编写从Matlab访问微阵列数据库的脚本。研究者也将在2005年6月和7月抽出相当一部分时间来做这个项目。软件开发：在这一年的过程中，确定了这个项目的软件将在两个平台上开发：Matlab和Davis。我们评估了几个可供选择的平台，包括genesspring和R，并认为这两个平台都不够灵活，无法支持这项工作所需的可视化类型。Matlab拥有庞大的复杂算法库，在其生物信息学工具箱和数据处理能力方面进行了重大改进。Matlab支持用户开发的gui（图形用户界面）。在Matlab中开发应用程序后，可以将其编译为不需要Matlab许可证的独立应用程序。Dragana Veljkovic致力于Matlab小波实现，并使用Matlab gui进行了一些原型开发。Davis（数据查看系统）是一个由研究者和她的学生用Java编写的数据可视化平台。Davis提供了Matlab中没有的数据处理基础设施。特别是，它支持多个同步视图，并可用于查看来自web的数据。它也是开发处理大型数据集的流算法的良好平台。在本资助年度，相当多的人员时间致力于稳定戴维斯平台，以实现未来的发展。研究人员在11月至4月期间对程序进行了重组，以便更容易添加新的可视化和新类型的数据。Cory Burkhardt为Davis重写了底层的同步和定时机制，并且已经开始记录程序，实现配置文件，并编写用户？年代的向导。Jason Edwards和James Packer重新设计了允许用户设置可视化参数（如颜色地图）的首选项。Egle Pilipaviciute致力于实现全局KL分解技术。研究者在Davis实现了一个小波功能，用于进行多分辨率可视化，以及将KL分解与小波相结合的技术。基线知识的获取：研究者继续获取背景知识。她定期参加生物信息学研讨会，这是一个研究生在计算生物信息学方面发表重要论文的每周会议。她在12月的RCMI会议上参加了微阵列技术的教程。她计划参加8月8日举行的IEEE计算系统生物信息学会议。2005年11月11日，以及与该会议相关的教程。她也是两名从事生物信息学研究的学生的博士论文委员会成员。今年，她与RCMI导师见了两次面：一次是2月份在德克萨斯州休斯顿举行的Coupled60会议，另一次是4月份在华盛顿特区举行的NSF计算神经科学合作研究会议。具体目标进展概述：具体目标1：开发微阵列数据集可视化分析的新方法。A.应用降维技术，如KL分解。研究者与生物学王玉峰共同指导mbr - rise学生Maribel Sanchez进行独立研究。使用Matlab和genesspring， Maribel应用KL分解来分析疟疾数据挑战数据集中的细胞周期，该数据集由加州大学旧金山分校的DeRisi实验室为CAMDA 2004（微阵列数据关键评估）竞赛制作。她发现KL分解捕获了细胞周期，并能够预测数据的阶段。开发和应用波浪分析的一般技术。除了A部分的细胞周期分析和作为Davis的一部分的一般波技术的发展之外，没有在这个特定目标上做任何工作。C.开发理解基因簇关系的视觉技术。CS博士生Robert Baltimore做了一个关于微阵列数据聚类可视化的独立研究。他特别关注了在受限方向上聚类微阵列数据的技术。开发微阵列数据的结构分析技术。MBRS/RISE的学生Magdaliz Gorritz做了一个独立的研究，她收集了大量的微阵列数据集以及其他信息的链接。她使用R程序对大量这些数据集进行同步分析。这些数据将用作正在开发的技术的测试数据。具体目标2：为微阵列数据集的多尺度分析和探索开发新的可视化技术。a .创建一个基于web的数据浏览器，用于在多个层次上导航微阵列数据集。在导航微阵列数据方面没有取得具体进展。然而，小波分析在戴维斯实现了多分辨率分析。B.将在线数据库与数据浏览器集成。2005年春，与CS硕士李钊共同指导了一项独立研究。对生物信息学感兴趣的学生。Li使用了COG数据库和论文中提供的补充数据。利用逻辑关系解读蛋白质网络组织？Bowers et al. （Science 306:2246-2249, 2004）。她实现了他们的算法，利用蛋白质三胞胎的系统发育特征来推断网络关系。我们计划使用这些逻辑关系来注释微阵列数据中的簇。这项工作还在进行中。C.使用3D技术和导航来探索微阵列数据。主人?他的毕业论文学生Mark Robinson继续致力于在3D中覆盖两个表面的技术的发展，以便比较标量数据集。主人?她的论文学生雷切尔·史密斯正在开发算法，并实现使用数据手套在3D数据中导航。马克和瑞秋都在进行用户研究，并且已经批准了人类受试者。形式。本科生Jason Johnson正在研究在Davis使用VTK（可视化工具包）进行3D可视化的可行性。我们已经在使用这些技术方面取得了进展，但还没有将它们应用于微阵列数据的阶段。新的合作：这项发展资助的另一个方面是在生物信息学领域形成新的合作。作为参与RCMI项目的直接结果，这位研究人员今年已经开始了三项新的研究合作：在猴子的运动皮层进行多电极记录。这些记录产生了大量具有波状活动的时空数据。他的数据将有助于研究数据处理和多分辨率问题。研究者已经为戴维斯可视化格式化了这些数据。他的一名本科荣誉论文学生道格·卢比奥（Doug Rubio）在12月拜访了这位研究员，并与他一起工作了两天，学习了波浪技术，并讨论了适用于这些数据的分析方法。Nicholas在2月份作为RCMI研讨会的演讲者来到这里，今年夏天，他们将通过Doug继续合作，分析数据中方向性的空间依赖性。2)科琳·维特？他是伯克利大学的博士后，研究t细胞发育过程中的细胞运动。她以前是理查德·拉巴伦的学生，理查德建议合作。研究者在Matlab中编写了一套分析工具，用于在双光子显微镜数据中查看细胞运动特征。这一经验将使研究人员能够协助其他研究人员使用双光子显微镜RCMI核心设备，该设备将于明年上线。3)马修·高文？UTSA RCMI项目主管，从事中枢呼吸化学接受研究。四月在休斯顿举行的RCMI会议上讨论了他的数据后，两位研究者意识到小波信号分析技术将适用于呼吸数据。两位调查人员将直接合作，并通过他们的研究生Vonnie Veit和Dragana Veljkovic进行合作。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. The specific aims of this grant are as stated in the original proposal. Since the grant was funded for 3 years rather than for 5 years, the investigator will focus more on development of specific analysis and visualization techniques and less on building general data-handling infrastructure than was originally proposed. Year 1 of the grant was to be devoted to acquisition of baseline knowledge and implementation of basic software infrastructure. Future work will continue as outlined in the proposal. Personnel: Funding for the grant did not officially arrive until November, 2004, delaying the initial hiring of staff until spring 2005. However, during the fall 2004 semester, the PI worked with two minority students on projects specifically addressing two aspects of the specific aims as described below. To compensate for lost time due to these delays, the investigator hired two half-time graduate research assistants (Egle Pilipaviciute and Dragana Veljkovic) and two undergraduate programmers (Jason Edwards and James Packer) in January of 2005. The full-time research software developer position was filled at the beginning of March, 2005 by Cory Burkhardt, a summa cum laude graduate of UTSA. The investigator also hired David Bigham, a recent UTSA graduate who is entering the CS Master?s program in the fall of 2005, to work for three months during the summer of 2005. David will be writing scripts to access microarray databases from Matlab. The investigator also will devote a significant portion of her time in June and July of 2005 to this project. Software development: During the course of this year, it was determined that software for this project will be developed on two platforms: Matlab and Davis. We evaluated several alternative platforms including GeneSpring and R and decided that neither of these platforms was flexible enough to support the types of visualizations needed for this work. Matlab, which has an enormous library of sophisticated algorithms, has undergone major improvements in its bioinformatics toolbox and data handling capabilities. Matlab supports user-developed GUIs (graphical user interfaces). After an application has been developed in Matlab, it can be compiled into a standalone application that does not require a Matlab license. Dragana Veljkovic worked on Matlab wavelet implementations and did some prototype development with Matlab GUIs. Davis (Data Viewing System) is a data visualization platform written in Java by the investigator and her students. Davis provides a data handling infrastructure that is not available in Matlab. In particular, it supports multiple simultaneous synchronized views and can be used to view data from the web. It is also a good platform for developing streaming algorithms for handling large data sets. Considerable personnel time during this grant year has been devoted to stabilizing the Davis platform to enable future development. The program was reorganized by the investigator during the period from November to April so that it would be easier to add new visualizations and new types of data. Cory Burkhardt rewrote the underlying synchronization and timing mechanisms for Davis and has begun documenting the program, implementing configuration profiles, and writing a user?s guide. Jason Edwards and James Packer reworked the preferences that allow the user to set visualization parameters such as color maps. Egle Pilipaviciute worked on the implementations of global KL decomposition techniques. The investigator implemented a wavelet capability in Davis for doing multi-resolution visualizations as well as techniques that combined KL decomposition with wavelets. Acquisition of baseline knowledge: The investigator continued to acquire background knowledge. She regularly attended the bioinformatics seminar, a weekly meeting in which graduate students present important papers in computational bioinformatics. She attended a tutorial in microarray technology at the RCMI meeting in December. She is planning to attend the IEEE Computational Systems Bioinformatics Conference, Aug. 8?11, 2005, along with tutorials associated with that meeting. She is also a member of doctoral dissertation committees of two students who are working in bioinformatics. She met with her RCMI mentor twice this year: at Coupled60, a conference that was held in Houston, TX in February and at the NSF Collaborative Research in Computational Neuroscience Meeting in April in Washington, DC. Summary of progress on specific aims: Specific aim 1: Develop new approaches for the visual analysis of microarray data sets. A. Apply dimension-reducing techniques such as KL decomposition. The investigator jointly supervised (with Yufeng Wang of Biology) MBRS-RISE student Maribel Sanchez in an independent study. Using Matlab and GeneSpring, Maribel applied KL decomposition to analyze cell cycles in the Malaria data challenge data set produced by the DeRisi lab at UC San Francisco for the CAMDA 2004 (Critical Assessment of Microarray Data) contest. She found that KL decomposition captured the cell cycle and was able to predict the phase of the data. B. Develop and apply general techniques for the analysis of waves. No work was done on this specific aim beyond the cell cycle analysis of part A and the development of general wave techniques as part of Davis. C. Develop visual techniques for understanding gene cluster relationships. CS PhD student Robert Baltimore did an independent study on visualization of clustering for microarray data. In particular, he looked at techniques for clustering microarray data in restricted directions. D. Develop techniques for structural analysis of microarray data. MBRS/RISE student Magdaliz Gorritz did an independent study in which she gathered a large number of microarray data sets as well as links to other information. She worked using the program R to run simultaneous analysis on a large number of these data sets. This data will be used as test data for the techniques being developed. Specific aim 2: Develop new visualization techniques for multi-scale analysis and exploration of microarray data sets. A. Create a web-based data browser for navigating microarray data sets at multiple levels. No specific progress was made on navigating microarray data. However, wavelet analysis for multi-resolution analysis was implemented in Davis. B. Integrate online databases with the data browser. In the spring of 2005, the investigator supervised an independent study with Li Zhao, a CS master?s student interested in bioinformatics. Li worked with the COG databases and the supplemental data provided in the paper ?Use of Logic Relationships to Decipher Protein Network Organization? by Bowers et al. (Science 306:2246-2249, 2004). She implemented their algorithm to use phylogenetic profiles of triplets of proteins to infer network relationships. We plan to use these logic relationships to annotate clusters in microarray data. This is work in progress. C. Use 3D technology and navigation to explore microarray data. Master?s thesis student Mark Robinson continued to work on the development of techniques for overlaying two surfaces in 3D in order to compare scalar data sets. Master?s thesis student Rachel Smith is developing algorithms and an implementation to use a data glove to navigate through data in 3D. Both Mark and Rachel are conducting user studies and have approved human subjects? forms. Undergraduate student Jason Johnson is investigating the feasibility of using VTK (Visualization toolkit) to do 3D visualization in Davis. We have made progress in using these technologies but are not at the stage of applying them to microarray data. New collaborations: Another aspect of this development grant is the formation of new collaborations in bioinformatics. The investiagor has started three new research collaborations this year as a direct result of her involvement with the RCMI program: 1) Nicholas Hatsopoulos ? University of Chicago, performs multi-electrode recordings in monkey motor cortex. These records produce large amounts of spatial-temporal data with wave-like activity. His data will be useful for looking at data handling and multi-resolution issues. The investigator has formatted this data for Davis visualization. One of his undergraduate honors thesis students, Doug Rubio, visited and worked with the investigator for two days in December to learn the wave techniques and to discuss what analysis will be applicable to this data. Nicholas came as an RCMI seminar speaker in February and the collaboration will continue this summer through Doug on an analysis of the spatial dependence of directionality in the data. 2) Colleen Witt ? a postdoctoral fellow from Berkeley, works on cell motility during T-cell development. She is a former student of Richard LaBaron, and Richard suggested the collaboration. The investigator has written a suite of analysis tools in Matlab to look at cell motility characteristics in two-photon microscopy data. This experience will allow the investigator to assist other researchers who will be using the two-photon microscopy RCMI core facility that should come on line next year. 3) Matthew Gdovin ? UTSA RCMI project director, works on central respiratory chemoreception. After a discussion of his data at the April RCMI meeting in Houston, the two investigators realized that the wavelet signal analysis techniques would be applicable to the respiration data. The two investigators will collaborate directly and through their graduate students, Vonnie Veit and Dragana Veljkovic.