High performance computing infrastructure for evolutionary biology, spatial ecology, and conservation biology

基因组学、生物信息学和地理信息学（地理数据的收集和分析）的快速发展彻底改变了我们的进化生态学、群落生态学和保护生物学领域，使我们能够收集前所未有的规模和范围的数据集。不幸的是，我们生成数据的能力已经超过了我们以具有成本效益的方式分析数据的能力;这对可以包括在学生研究项目中的分析的类型和质量、研究生的完成时间以及我们吸引优秀学生和教师的能力产生了深远的影响。对于可并行化的密集型计算机任务，正如我们的大多数分析一样，一个明显的解决方案是使用谷歌或亚马逊等商业提供商提供的云计算，或者使用加拿大计算资源。前者对于像我们这样的NSERC支持的主要研究者来说太昂贵了，我们是通过评估小组1503（生态学和进化）资助的，并且每年指导许多学生和博士后。加拿大也出现了问题。首先，像我们的学生这样的用户通常被降级到最低优先级，这可能意味着作业启动的明显延迟。其次，虽然有赠款（资源分配竞赛）支付研究人员访问加拿大计算集群，但这些要求个人PI超过50核心年的预计CPU使用量，这是大型医疗团体的典型数量，但不是绝大多数生态学和进化研究人员。为了解决我们研究中的这一迫切需求，我们建议在我们的高级计算中心（CAC）的协助下，为女王学院的生物学系创建一个专门的生物学子集群。子集群将有80个核心，768 GB CPU RAM和64 GB GPU RAM，96 TB可访问的网络附加存储，额外的80 TB长期磁带存储，并且没有用户费用。我们使用超过25种不同的软件包或工作流程，这种本地基础设施加上CAC的支持，使我们能够最大限度地灵活调整软件，以最有效地使用，并为HQP的个性化需求开发特定的管道。他们的实物支持将包括每学期为所有人员提供有针对性的培训，以最大限度地提高他们利用这一新的强大资源的能力。这个新的计算机基础设施将服务于我们的计算需求在未来6年，并将直接支持所有五个PI的研究，至少49个本科生荣誉学生项目，45个硕士和博士生，和5个博士后研究员通过我们自己的研究计划。虽然我们的HQP将有优先权，但我们打算与生物学的所有教师（>30）和学生（约120名研究生）分享这种新的计算能力。我们将利用这一基础设施，以吸引不同的新教师和最高水平的学生，并提高我们部门的所有成员的学习和研究环境。**

Rapid advances in genomics, bioinformatics, and geomatics (collection and analysis of geographic data) have revolutionized our fields of evolutionary ecology, community ecology, and conservation biology, allowing us to gather datasets of unprecedented size and scope. Unfortunately, our ability to generate data has outstripped our ability to analyse them in a cost-effective manner; this is having profound consequences for the types and quality of analyses that can be included in student research projects, for finishing times for graduate students, and for our ability to attract to calibre students and faculty. An obvious solution to intensive computer tasks that are parallelizable, as most of our analyses are, is to use cloud computing provided by commercial providers like Google or Amazon, or to use Compute Canada resources. The former are simply too costly for NSERC-supported principal investigators like us who are funded through Evaluation Group 1503 (Ecology and Evolution), and who mentor many students and postdocs annually. Compute Canada too presents issues. First, users like our students are typically relegated to lowest priority and this can mean pronounced delays for jobs to start. Second, while there are grants (Resource Allocation Competitions) to pay for researchers to access Compute Canada clusters, these require individual PIs to exceed 50 core years of projected CPU usage an amount this is typical of large medical groups but not of the vast majority of ecology and evolution researchers. To address this dire need in our research, we propose to create a dedicated Biology sub-cluster for the Department of Biology at Queen's, facilitated by our Centre for Advanced Computing (CAC). The sub-cluster would have 80 cores with 768GB CPU RAM and 64GB GPU RAM, 96TB of accessible Network Attached Storage, an additional 80TB of long-term tape storage, and no user fees. We use over 25 different software packages or workflows and this local infrastructure coupled with the support from CAC allows us maximum flexibility to adapt software for most efficient use and develop specific pipelines for individual needs of our HQP. Their in-kind support will include tailored training for all personnel each term to maximize their ability to take advantage of this new and powerful resource. This new computer infrastructure will serve our computational needs for the next 6 years, and will directly support the research of all five PIs, at least 49 undergraduate honours student projects, 45 MSc and PhD students, and 5 postdoctoral fellows through our own research programs. While our HQP would have priority, we intend to share this new computational capacity with all faculty (>30) and students (~120 graduate students) in Biology. We will use this infrastructure to attract diverse new faculty and students of the highest calibre, and to enhance the learning and research environment for all members of our department. **