In the past, my research has focused on understanding how changes in the regulatory sequences of DNA contribute to cancer development and progression. To this end, I developed novel algorithms for investigating cancer expression data in the areas of differential coexpression and subspace clustering. Complementary to these approaches was the curation and collection of known regulatory sequences and mutations. To this end, I developed ORegAnno (www.oreganno.org), an open access database and curation system for regulatory sequences and polymorphisms. This system has captured the experimental results of hundred of studies of genetic and epigenetic regulation. My research continues to focus on the development of open access and open source resources for cancer expression analysis.
My research interests are best described as cancer informatics and clinical statistics. Specifically, I use computational methods for the analysis of large cancer datasets at the molecular and genomic level. I have worked on the identification of molecular markers at the DNA, RNA and protein level that are useful for diagnosis and prognosis of cancer. Using bioinformatics methods, I have conducted extensive meta-analyses of expression profiling studies for thyroid cancer. This allowed the identification and ranking of high-confidence differentially expressed genes between cancer and non-cancer states. Such candidates were then validated by immunohistochemistry and tissue microarray analysis on a cohort of over two-hundred human thyroid tumour patient samples. This allowed the development of a panel of markers that can classify malignant versus benign thyroid neoplasms with high accuracy (>90%) and thus has potential utility as a diagnostic tool for thyroid cancer. The approach was also extended to colon and rectal cancer.
More recently, my research focus has shifted to the analysis of next-generation sequencing data and clinical statistics in breast cancer, a disease with strong personal interest. I used similar meta-analysis methods as above to develop useful predictors for recurrence risk in different breast cancer subtypes. I have also helped to develop the Alexa-seq platform (www.alexaplatform.org) for analysis and visualization of RNA-seq data in terms of expression, differential expression and alternative expression. I applied this approach to a number of cell line and primary tumor datasets with the goal of characterizing their transcriptional profiles, with special focus on known and novel alternative splice forms. In addition to RNA-seq data, I worked on methods for integrating methylation, reverse phase protein array, exome-sequencing, and other data types with the goal of developing sophisticated ‘omic predictors of drug response in breast cancer. More detailed descriptions of my research projects can be found on my research webpage (www.obigriffith.org).
Currently I am engaged in a number of large-scale tumor sequencing projects on liver and breast cancer, investigating primary, relapse and drug resistant tumors. To this end I am working with others to develop an end-to-end pipeline for clinical cancer sequencing which automates state-of-the-art methods for sequence alignment, somatic variation detection, RNA sequence analysis, and the integration of these data types into a user-friendly report of the most clinically relevant genome and transcriptome changes in a tumor or group of tumors (github.com/genome/gms/wiki). To aid in this effort, I develop and maintain resources for interrogating the druggable genome (www.dgidb.org) and crowd-sourcing the curation of clinically actionable variants in cancer (www.civicdb.org).
In addition to my basic and clinical research interests, I am also passionate about the scholarship of teaching and learning. I have made substantial contributions to the training and education of tomorrow's bioinformaticians and biostatisticians through my involvement in CBW and CSHL workshops and BioStars bioinformatics forum. I enjoy lecturing, creating course materials and developing active teaching methods that promote effective learning. I am currently developing a bioinformatics and clinical informatics training program that takes a practical, hands-on approach to cancer genome analysis.