The goal of the Biostatistics Core is to ensure appropriate biostatistical support for cancer-related research at UCSF through a recognized and accessible central resource. The intent is to support in vitro and animal research as well as clinical research (defined as studies of disease etiology, both genetic and environmental, and methods of disease prevention, as well as studies of disease interventions, with either curative or palliative intent). Support is limited to cancer-related research, preferentially for Cancer Center members. Priority is given to currently funded research; among unfunded research projects (including institutional clinical trials and pilot projects), prioritization is set in collaboration with Cancer Center program leaders. As time permits, consultation is available to non-Cancer-Center members through a recharge mechanism.

Biostatistics personnel will participate in the writing of grant proposals and help in the planning of preliminary studies. Assistance in the quality control of data and the analysis of interim and final study results is also available. Additionally, support is provided for writing reports for publication and presentation.

Establish Account  |  Contacts >
Recharge Rates >

Services

General biostatistical support in study design
While the Core's first priority is clinical trials, issues such as sample size, controls, and selection of endpoints are of importance in all cancer research. Involvement of statisticians in the planning phase helps to ensure that studies can generate maximum relevant information. Biostatistics personnel provide input into study design as members of Cancer Center Programs and Clinical-Site or Protocol Review committees, in addition to responses to direct requests from individual members.

Guidance on study conduct
Core personnel advise on data definitions, database structure, and resolution of exception questions as they occur during the conduct of studies. Routine interaction between statisticians and study PIs allows for effective input as such questions arise; through ongoing availability to the study team, Core members may be able to identify potential problems that have otherwise unrecognized statistical implications. Core personnel also participate in the formal activities of the Cancer Center's Data Monitoring Committee.

Support in analysis and reporting of study results
Depending on the project, the statistician may analyze actual data. Alternatively, especially in the case of lab studies, the statistician may identify the appropriate analysis, write the statistical methods, and review study conclusions for consistency with the statistical analyses. The approach depends on the complexity of the analyses and availability of the research group to provide the needed support. Core members can develop and evaluate specialized statistical analysis programs as needed in support of these activities.

The Core provides data analysis support of microarray data via a recharge mechanism. While projects may require specific analyses, standardized approaches are also available. The statistician will perform quality checks appropriate for the data type and report the results to the investigator if quality appears to be unsatisfactory. Finding differentially expressed genes or clones with differential copy number is another common procedure in microarray studies. The statistician will apply unsupervised clustering procedures to the data, and for larger datasets where classification is of interest, may use standard discrimination methods to predict the class of interest. Overall, the experience of Core personnel with high-dimensional genomics data allows them to help investigators to formulate testable statistical hypotheses as well as to choose appropriate bioinformatics tools.

Microarray analysis work request
Microarray analysis consultation up to one hour per project are provided at no cost. Consultation beyond one hour, and all data analyses, are handled through a recharge mechanism. The appropriate form is available for download here.

Education

Faculty and staff are available to give short presentations on topics such as:

• Statistical issues in the design/conduct of clinical trials
• Sample size determination
• Interim analyses
• Multiple inference
• Why multivariate analysis?
• Role of logistic model in studies of LD50s
• When is a "paired" analysis appropriate?
• Determining quality of your microarray data
• Finding differentially expressed genes
• What clustering can and can not do for you
• Building a predictor
• Introduction to R statistical package

Other resources:
Dept. of Epidemiology & Biostatistics:
http://www.biostat.ucsf.edu/courses_index.html
http://www.biostat.ucsf.edu/cbmb/bmi209/

Statistical Methods in Radiology Research:
http://www.cmfi.ucsf.edu/research/biostatistics.shtml