You are here
Biostatistics and Computational Biology Core
The mission of the Biostatistics and Computational Biology Core (BCB) is to provide quantitative support for intramural collaboration and coordination of all AIDS-related research activities at Duke and its partner institutions. By providing quantitative expertise and developing shared computational tools, we aim to enhance the value of the services provided by other CFAR Cores, and to increase the scientific impact of research done by CFAR investigators.
The core mission is executed via the following specific aims:
- Support: Assist CFAR investigators with their computational and biostatistics needs for posters, papers and proposals
- Teach: Upgrade quantitative analysis level for HIV/AIDS research through education and mentoring
- Innovate: Develop useful methods and software for the HIV/AIDS research community
Support for grant applications by CFAR principal investigators
In clinical trials, power calculation for sample size ensures that we have sufficient sample size for achieving a desired power for detecting a clinically meaningful treatment effect at a pre-specified type I error rate. Power calculation for sample size is critical especially when only limited budget and/or resources are available. The BCB Core will assist CFAR PIs in (i) power calculation for sample size estimation and/or justification, (ii) evaluation the merits and disadvantages of alternative designs, and (iii) preparation of statistical section for inclusion in the grant application for statistical/scientific validity of the grant application.
Support for clinical studies sponsored by CFAR principal investigators
The BCB Core will provide data management (including data quality and verification), statistical programming, and data analysis support to clinical studies sponsored by CFAR investigators to ensure unbiased (accurate) and fair (reliable) assessment of test treatments under investigation. In addition, the BCB Core will work closely with the investigators for interpreting the analysis results and preparing the manuscripts for publication consideration in leading medical journals.
Training on statistical issues commonly encountered in AIDS clinical research
In clinical trials, some critical statistical/scientific issues such as the selection of non-inferiority margin in non-inferiority (active control) trials, appropriate methods for missing data imputation, the establishment of predictive model using genomic markers, and advantages of clinical trial simulation are commonly encountered in AIDS clinical research. BCB will identify specific issues/topics that are of particular to CFAR investigators and then provide training such as seminars, tutorials, short courses, or workshops to CFAR investigators at regulatory basis.
Design and analysis of AIDS clinical research, especially on the potential use of adaptive clinical trial designs and statistical design for small scale exploratory studies
In recent years, the use of innovative adaptive methods in clinical trial has become very popular due to its flexibility and efficiency for identifying any signal, pattern, and trend of clinical benefits of test treatments under investigation. However, the quality, validity, and integrity of data collected from clinical studies utilizing adaptive clinical trial designs are of great concern to principal investigators and regulatory agencies. BCB Core will provide statistical consultation and support to clinical studies intend to use adaptive designs such as group sequential design, adaptive dose finding design, phase I/II seamless adaptive design, biomarker-adaptive design (target clinical trials), and other innovative adaptive clinical trial designs that are commonly used in AIDS clinical research to make sure the quality, validity and integrity of the intended clinical trials utilizing innovative adaptive clinical trial designs.
Perhaps a more pressing problem for typical CFAR investigators is how to design effective small-scale exploratory studies on a shoestring budget. Good statistical planning is critical for such studies, since small poorly designed studies may provide no useful information at all. We are interested in combining exploratory HIV/AIDS studies with clinical trial simulations in order to provide insight into the optimal design and strategies for using these small scale studies to launch randomized clinical trials.
Software and algorithms for analysis of immunological assays
Sometimes, vendor-provided software for the analysis of data from immunological assays perform poorly on specific data sets or are simply not available (for newly developed assays). We are happy to work with CFAR investigators to help find or develop the appropriate algorithms or software packages to analyze and summarize immunological assay data. For example, we have previously developed software and algorithms for the analysis of iTopia, Luminex and polychromatic flow cytometry assay data in collaboration with CFAR investigators.
Mathematical and statistical modeling of host-pathogen dynamics
We are happy to work with CFAR investigators who wish to develop and calibrate mathematical models of some aspect of host, pathogen or host-pathogen dynamics. Such models are typically expressed as systems of nonlinear difference equations or ordinary differential equations, and can provide insight into biological mechanisms typically not available with standard statistical approaches. For example, the effect of combination antiviral therapy was initially evaluated with the help of such models for viral dynamics.
Workshops on practical computing for research scientists
The BCB core runs annual workshops on practical computing for CFAR research scientists. The objective of these workshops is to increase the productivity of researchers by the use of software tools such as the Unix shell, text editors, databases, Python and R to facilitate data management, manipulation, analysis and reporting. These workshops are intended to convert novices into competent (not expert) users of powerful computational tools.
Help with statistical plan for CFAR small grant applications
The BCB core helps with the experimental design and analysis plan for many CFAR small grant applicants. This is a free service, and the Core also provides support for data analysis upon request if the grant application is successful. Some examples of small grants where a critical statistical review was provided are:
- Susanna Naggie: The role of antiretroviral-induced dyslipidemia in lipid-HCV interactions in HIV/HCV co-infected patients: An assessment of the effect on HCV viral integration
- Wanda Lakey: From wasting to obesity - Antiretroviral therapy and weight gain in HIV-infected patients
- Giny Fouda: Specificity and function of maternally-acquired and vaccine-induced antibody responses in HIV-exposed infacnts
- Christy Kaiser: Comparison of screening rates for coronary artery disease in HIV-infected and HIV-uninfected patients
- Maria Biasi: The ability of integrase-defective lentiviral vector (IDLVs) to elicit durable antibody responses in systemic and mucosal compartments
- Dorothy Dow: Establishing mental health needs in HIV-positive adolescents in Tanzania
Data analysis for prospective cohort studies of fungal, bacterial and viral infections in sub-Saharan Africa
The BCB core performed data analysis for a series of prospective cohort studies of infectious diseases in Moshi, Tanzania in collaboration with John Crump and John Bartlett. These studies have been described in the following publications.
- Crump, J. A., Ramadhani, H. O., Morrissey, A. B., Saganda, W., Mwako, M. S., Yang, L.Y., Chow, S. C., Reyburn, H., Njau, B. N., Shaw, A.V., Bartlett, J. A., and Maro, V. P. (2012). Bacteremic disseminated tuberculosis in sub-Saharan Africa: a prospective cohort study. Clinical Infectious Diseases: 55:242–250. [SCI]
- Crump, J.A., Ramadhani, H.O., Morrissey, A.B., Saganda, W., Mwako, M.S., Yang, L.Y., Chow, S.C., Njau, B.N., Mushi, G.S., Maro, V.P., Reller, L.B., Bartlett, J.A. (2012). Bacteremic disseminated tuberculosis in sub-Saharan Africa: a prospective cohort study. Clinical Infectious Diseases, 55, 242–250.
- Crump, J.A., Ramadhani, H.O., Morrissey, A.B., Saganda, W., Mwako, M.S., Yang, L.Y., Chow, S.C., Morpeth, S.C., Reyburn, H., Njau, D.N., Shaw, A.V., Bartlett, J.A., and Maro, V.P. (2011). Invasive bacterial and fungal infections among hospitalized HIV-infected and HIV-uninfected adults and adolescents in northern Tanzania. Clinical Infectious Diseases, 52, 341–348.
- Crump, J.A., Ramadhani, H.O., Morrissey, A.B., Msuya, L.J., Yang, L.Y., Chow, S.C., Morpeth, S.C., Reyburn, H., Njau, B.N., Shaw, A.V., Diefenthal, H.C., Bartlett, J.A., Shao, J.F., Schimana, W., Cunningham, C.K., and Kinabo, G.D.(2011). Invasive bacterial and fungal infections among hospitalized HIV-infected and HIV-uninfected children and infants in northern Tanzania. Trop Med Int Health, 16, 830–837.
RCT Of An Integrated Treatment Of Persons With Co-Occurring HCV And Alcohol Abuse
Alcohol abstinence greatly improves outcomes for HCV-nfected patients. A 6-month treatment model for alcohol abstinence in these patients was proposed by Rae Jean Proeschold-Bell, and the BCB core helped design the statistical analysis plan for this randomized control trial which was recently awarded an R01 by NIH on resubmission.
Methodological studies on adaptive trial design
In recent years, the use of adaptive design methods in clinical research and development based on accrued data has become very popular due to its flexibility and efficiency. An adaptive design allows modifications made to trial and/or statistical procedures of ongoing clinical trials. However, it is a concern that the actual patient population after the adaptations could deviate from the originally target patient population and consequently the overall type I error (to erroneously claim efficacy for an infective drug) rate may not be controlled. In addition, major adaptations of trial and/or statistical procedures of on-going trials may result in a totally different trial that is unable to address the scientific/medical questions the trial intends to answer. Some recent publications from the BCB core on adaptive trial design include:
- Chow SC. Adaptive clinical trial design. Annu Rev Med. 2014 Jan 14;65:405–15. doi: 10.1146/annurev-med–092012–112310. PMID: 24422576
- Chow SC and Chiu ST (2013). Sample size and data monitoring for clinical trials with extremely low incidence rate. Therapeutic Innovation and Regulatory Science, 47:438–446.
- Chow SC, Chiu ST (2013) A Note on Design and Analysis of Clinical Trials. Drug Des 2:102. doi: 10.4172/2169–0138.1000102
Statistical mixture modeling for cell subtype identification in flow cytometry
Statistical mixture modeling provides an opportunity for automated identification and resolution of cell subtypes in flow cytometric data, and may reduce the subjectivity inherent in manual gating procedures. The BCB core, working with the Flow Cytometry Core, published the first paper on the use of multivariate Gaussian mixture models for cell subset identification from flow cytometry data in 2008, and this continues to be an active area of research. Some relevant publications are:
- Chan C, Feng F, Ottinger J, Foster D, West M and Kepler TB, Statistical mixture modeling for cell subtype identification in flow cytometry, Cytometry A, (2008), 73A:693–701
- Frelinger, J, Ottinger J, Chan C, Modeling flow cytometry data for cancer vaccine immune monitoring, Cancer Immunology Immunotherapy, (2010), 59:1435–41.
- Chan C, Lin L, Frelinger J, Hérbert V, Gagnon D, Landry C, Sékaly RP, Enzor J, Staats J, Weinhold KJ, Jaimes M, West M. Optimization of a highly standardized carboxyfluores- cein succinimidyl ester flow cytometry panel and gating strategy design using discriminative information measure evaluation, Cytometry A, (2010), 77:1126–36.
- Cron A, Gouttefangeas C, Frelinger J, Lin L, Singh SK, Britten CM, Welters MJP, van der Burg SH, West M, Chan C,. (2013) Hierarchical Modeling for Rare Event Detection and Cell Subset Alignment across Flow Cytometry Samples. PLoS Computational Biology 9(7): e1003130. doi:10.1371/journal.pcbi.100313.
Core Staff Contact
Cliburn Chan MBBS Ph.D. (Core director/Computational Biology)
Immune response models, assay analysis and immune profiling
Shein-Chung Chow Ph.D. (Core co-director/Biostatistics)
Clinical trials, pharmaceutical statistics and biomarker discovery
Greg Samsa Ph.D.
Statistical education (DGS for Biostatistics department), analysis of surveys
Darongsae Kwon Ph.D.
Shih-Ting Chiu Ph.D.
Scott White B.Sc.