The Biomedical Statistics Research Group (BSR), located in the Li Ka Shing building, is pleased to announce an opportunity for biology labs to find help with their data analysis needs by engaging in collaboration with statisticians in our group. Graduate students and post-docs from the BSR will collaborate with biology labs that are in need of assistance with the statistical analysis of their data. We are currently soliciting proposals from members of the UC Berkeley biology community, and will pair up the chosen proposals with students and post-docs from the BSR.

To learn more about how we envision that these directed collaborations will be structured and whether you could benefit by this type of collaboration, you can read more about the details below. If you have more specific questions not addressed by this website, email the faculty member overseeing these collaborations: Elizabeth Purdom (epurdom@stat.berkeley.edu).

In addition to this resource, the Statistics Department also runs a consulting service for the entire campus community during the academic semester. If you merely need to sit down and speak with someone about how you should best analyze your data on your own, this is a good resource. You can schedule a session with a graduate student in the Statistics Department and receive advice regarding your statistical analysis. Please see their website, http://www.stat.berkeley.edu/consulting, for more details.

The Nature of the Directed Collaborations

The goal of the directed collaborations is to engage our students and post-docs in active collaborations with biology labs on campus. This is a valuable experience for our students and post-docs, and we hope to continue to strengthen our ties with the UC Berkeley community. While we would love for some of these directed projects to take off into long-term relationships with the students or our faculty, many of these collaborations will not be the primary research focus of our students or post-docs, so they are designed to be of limited scope.

Our members' expertise is in statistical analysis and design, particularly as applied to the field of bioinformatics. Some common areas of analysis that would be well suited for our consultants are: designing an experiment, performing hypothesis tests, predicting outcomes, and finding clusters or patterns in data. Common types of experiments include microarray or sequencing experiments. These are only examples, and we are glad to work with groups on statistical problems that do not fall neatly into any of our backgrounds. Proposals are not limited to any particular question or topic; generally we are glad to work on any interesting question posed by the data. While our consultants often have experience in processing data, computer coding, and other computationally intensive tasks and maybe able to advise on these issues, the consultant's contribution should be centered on the statistical analysis.

The Structure of the Directed Collaborations

Our consultant will meet with the lab member(s) and discuss the project in more detail. At that time, they will develop a plan for how to best use the consultant's time. Consultants are committing to the BSR the equivalent of three hours per week during the semester for the project to which they are assigned (if needed), though the hours do not have to be spread evenly throughout the semester, depending on the lab and the consultant's needs. The initial plan will outline what is hoped to be accomplished during this collaboration and an initial schedule.

During this process, we encourage labs to think about the best use of the consultant's time and expertise. Consultants will often need an initial period of time to explore the data and come to understand the problem before they are able to give a reasonable estimate of the complexity of the question. Given the constraints of the consultant's involvement on the project, the lab should also consider how they will be able to pick up from where the consultant leaves off.

The collaboration's progress will be monitored by the faculty in charge via meetings with the faculty and the consultant. The consultant will be expected to create a final writeup such that could be adapted for inclusion in a research article by the lab, and the consultant is also expected to put the numerical results in a format that the lab could use to do further analysis on their own.

We assume that appropriate credit will be given to the consultant in any research publications that use the results of the collaboration.

Submitting a proposal

Interested members of the LKS community should submit a brief description of their project, their data, and the statistical analysis for which they would like assistance to epurdom@stat.berkeley.edu .

This is not a formal process, and there are no specific requirements. The goal is simply to have a reasonable summary understanding of the project for which the consultant is desired. The proposal should be comprehensible to the biostatisticians who will be reading it. The most important part of the proposal is to describe data that the lab has (or plans to have) and the question that needs to be addressed. If the data does not yet exist or has not yet been processed, please be clear on when the data that needs to be analyzed will be ready.

We will give priority to proposals from anyone within the LKS or to research related to the LKS core research areas. Graduate students, post-docs, or faculty may submit a proposal. Please be clear in the proposal as to who will be the main contact for the collaboration, their position and contact information, and from which lab the proposal is coming (there are no limits on how many proposals come from different members of a lab, but they should represent different projects).

Who are we?

The Biomedical Statistics Research Group is a group of faculty from the Division of Biostatistics and the Department of Statistics that work on statistical applications to biomedical research.
We are located in 344 Li Ka Shing Center.

Our faculty members are

Peter Bickel
Sandrine Dudoit
Steve Evans
Haiyan Huang
Alan Hubbard
Mark van der Laan
Elizabeth Purdom
Terry Speed

Back to top