P30 Vision Research Core
Biostatistics for Eye and Vision Research
Laurel Stell, PhD, is in the Department of Biomedical Data Science, but Ophthalmology is funding part of her salary to provide biostatistical support for its clinical and laboratory research. She has extensive experience in statistical modelling (developing predictors or classifiers), inference, and multiple hypothesis testing; but she can assist in many types of analyses. She is particularly adept at manipulating and visualizing data to discover general patterns and highlight interesting features and relationships—or perhaps errors. Furthermore, she can help guide your research approach before you even have data.
To arrange a meeting to discuss how Laurel can help with your project, please fill in the consultation request form. Due to her other commitments outside of Ophthalmology, it is best to allow plenty of time for her to fit your analysis into her queue. You should also consider whether any of the resources listed below are more appropriate, depending upon your needs and affiliations. If you have funding or would like to include a biostatician on a grant, that will also affect your options.
Other Biostatistics Resources
The Quantitative Sciences Unit (QSU) is a collaborative statistics unit in the Biomedical Informatics Research (BMIR) Division in the Department of Medicine. They do not have an arrangement with Ophthalmology, but they do have arrangements with Stanford Cancer Institute (SCI), Child Health Research Institute (CHRI, including applicants for their grants), and several other centers do. For an up-to-date list, see the QSU Project Initiation Form (about a third of the way down).
The Department of Biomedical Data Science (DBDS) provides free time-limited consultation to investigators in the School of Medicine. It also collaborates on longer term projects with other departments. In particular, if you are designing a clinical trial, you should request a consultation with DBDS. For more information about DBDS services, click here.
Data Studio is a collaboration between Spectrum (The Stanford Center for Clinical and Translational Research and Education) and DBDS. Its fundamental goal is to foster dialog between statisticians/data scientists and researchers in clinics and laboratories. Data Studio meets weekly during the fall, winter, and spring quarters. One session per month is drop-in consulting. Most sessions, however, consist of an extensive and in-depth consultation for a Medical School researcher.
Bioinformatics and Omics Resources
Bogdan Tanasa, MD, PhD, collaborates with the faculty in the Department of Ophthalmology on the bioinformatics analysis of the genomics datasets. His expertise is in integrative genomics and in the computational analysis of a large variety of HT-seq technologies (including ChIP-seq, RNA-seq, ATAC-seq, GRO-seq/PRO-seq, 4C, 5C, HiC, WES (whole exome sequencing), WGS (whole genome sequencing). His Stanford email account is bogdan.tanasa.
More About Bioinformatic Analysis
As with "biostatistics", the term "bioinformatics" is very broad, encompassing many different areas of expertise. The remainder of this section gives a high level overview of some of the relevant domains and resources.
The first step in omics research is to collect samples and perform laboratory work. This might involve sequencing, which can be done by the Genome Sequencing Service Center (GSSC) or the Stanford Functional Genomics Facility (SFGF) (which also provides microarray services). The resulting FASTQ files must be aligned to a reference genome and then quantified in some way. This is one of Dr. Tanasa's areas of expertise, and the Genetics Bioinformatics Service Center (GBSC) also does this. These Stanford service centers and facilities are fee-based, whereas Dr. Tanasa is employed by Ophthalmology.
Once measurements have been made by whatever means, the data must be visualized and analyzed to gain insight. Common goals are to discover associations between genomic variants and phenotypes or to identify genes that are differentially expressed between groups of samples. There are many software packages and Web resources to help you do this yourself. Dr. Tanasa is also knowledgeable about these pipelines, and the GBSC services extend to this area as well. If your data require adapting the pipeline or interpreting unusual results, additional statistical expertise may also be beneficial. Some examples of the complications are:
- Are there batch effects? Are there effects from other factors, including hidden factors? How should all these factors be taken into account without biasing the experiment?
- When should data be log-transformed? Should the statistical model include an interaction term or random effects? Which machine learning algorithm is appropriate?
- Why don't the largest estimates of effect size or fold change have the smallest p-values? Which is more important? How do I shrink fold change estimates so that they agree better with the p-values?
- How should I correct for testing a huge number of hypotheses? What if I have many phenotypes as well as many genes or variants? What if I am looking for differential expression between more than two groups?
Dr. Stell and Dr. Tanasa can collaborate with you and with each other to ensure that you have the necessary expertise for your particular analysis. Dr. Stell also has extensive experience in statistical modelling, which is relevant, for example, in developing biomarker signatures.
Once the initial analysis identifies a set of genes or proteins of interest, further exploration of that set is often necessary in order to gain insights. This typically involves incorporating annotation into the analysis, as in gene set enrichment analysis for example. Dr. Tanasa can provide more information about these techniques.
Data Management and Database Repositories
There are many SOM resources to help with data access. If you need to create and manage your own database, you might consider using REDCap, which requires funding for certain services, or the more advanced services provided by the Data Coordinating Center (DCC), which is a service center.
Stanford also has access to many existing databases containing data from a wide array of sources such as:
- STARR (formerly known as STRIDE)
- The Data Center at the Center for Population Health Sciences
- For SCI members, the Stanford Cancer Institute Research Database (SCIRDB) has a rich set of data that integrates many resources.
The menu at the SOM IT Web page lists some additional databases.
This is an initial start at listing SOM resources that might be useful to Ophthalmology researchers. It will be updated based on feedback and as the SOM updates its Web pages.
Some additional "meta" resources are:
- SOM list of core research facilities. Some of the links on this page are broken, but you can Google the name of the facility to find its new page.
- Spectrum, Stanford's Clinical and Translational Science Award (CTSA)-supported research hub. Spectrum is an independent, interdisciplinary center that facilitates clinical and translational research across the university. Its Web site has resources on informatics, research IT, study design, regulatory affairs, and much more.
- DoResearch is a Stanford University knowledge base, oriented primarily toward policy matters.