The Bioinformatics and Biostatistics Core at Joslin Diabetes Center offers support for data-driven projects related to basic, clinical and translational research, with a particular emphasis on diabetes. The core aims to ensure that researchers take advantage of the most modern and robust methods available in the field of Bioinformatics and Biostatistics
We offer analysis of all high-throughput data types, including: gene expression data from microarrays, qPCR, and RNA-seq; proteomics, phosphoproteomics, metabolomics and lipidomics from mass spectrometry and SomaLogic; phylogenetic and metagenomics analysis of DNA sequences; and DNA methylation sequencing data.
The typical pipeline includes: normalization, quality control, Principal Component Analysis (PCA), differential expression, pathway analysis, and visualization. Additional features: sample size and power calculations for high-throughput studies; state-of-the-art reproducible workflows; analysis and meta-analysis of public data; novel network analysis methods; integration of multiple data types, including clinical covariates; causal inference testing (AKA mediation analysis); global metabolic flux inference from Seahorse assays; an in-house searchable gene expression database with >75 studies (output from two studies for searching a gene of interest is shown in Figure 2) Bioinformatics services are provided by Drs. Jonathan Dreyfuss and Hui Pan.
We offer analysis of data from clinical, basic, translational and epidemiologic research often including comparisons of group means (e.g. t-test, ANOVA, non-parametric tests), measures of association (e.g. correlation, regression), time-to-event analyses (e.g. survival analysis, Cox regression), and mixed models/repeated measures approaches. Projects requiring modern approaches to dealing with missing data, simulation, and machine learning applications may also be accommodated.
During the initial consultation we typically determine the nature of the variables to be analyzed and the experimental design and questions of interest, reaching a consensus with the client regarding specification of tables and figures to be produced with an initial analysis plan to follow. Specifically we can assist with or provide instruction in:
• Sample size and power calculations for clinical, basic science, translational, and observational research
• Regression and multiple regression
• Logistic regression, ordinal/polytomous regression
• Non-parametric analyses
• Mixed models including Repeated Measures
• Discrete and continuous time-series
• Survival analysis, Cox proportional hazards regression (Figure 3)
• Modern approaches to treating missing data
• Machine learning approaches
• Statistical design of experiments
• Preparation of Methods, Statistical analysis plan for manuscripts and grant applications
Biostatistics services are provided by Dr. David Pober.