Data Analysis
Data analysis projects may involve data processing, application of
standard or novel techniques, and production of a report. Reports are
typically created programmatically, where outputs from the code are
directly incorporated into the document, to enable reproducibility of
the full analysis.
Example data analysis projects include:
Predictive modelling using proteomics data
Building a classification model using penalized logistic regression
to evaluate the additional predictive value of proteomics data over
clinical predictors.
Analysis of timecourse microarray experiment
Exploratory analysis of sample similarities using principal components
analysis. Assessment of significant genes using generalized least
squares model to account for correlation over time (using
the limma package). Use gene set enrichment analysis to
identify significant gene sets.
Mediation analysis of an intervention study
Applying the R package mediation to investigate
the mediating effect of a individual's belief regarding
treatment on the effect of treatment in a clinical trial
for restless leg syndrome.
Methodology Studies
The objective of a methodology study may be to research the state of the
art in a particular area, to propose a novel approach or to evaluate
competing methods both in theory and practice.
Example methodology projects include:
Methods to identify synergism/antagonism
Research and development of procedures for evaluating the significance of
departures from an additive model for compound interactions.
Exon array software evaluation
Evaluating JETTA package for the analysis of Human Exon 1.0 and
Human Transcriptome 2.0 arrays. Providing analysis templates, using
customised R functions.
Gene set enrichment analysis review
Reviewing current methodology for gene set enrichment analysis and
identifying methods that may provide some benefit over the method
currently in use by the client. Applying selected methods to internal
data sets to assess performance in practice and investigate differences
between the methods.
R Programming
An R programming project may involve writing a script to perform
specified tasks using existing R functions, implementing a novel method
in R, or developing a custom R package.
Example programming projects include:
R function to fit Bayesian Model
Coding block-update Gibbs sampler in R for Bayesian hierarchical model
that client was unable to fit using WinBUGS. Provision of example
analysis using client's data on retail sales.
R package for Dose Response Modelling
Development of custom R package from prototype code for mixed effects
dose response modelling. Follow-on “Generation II” package combining
stages of analysis.
Demonstration scripts for R novice
Writing R scripts to demonstrate clustering, logistic regression and
Cox regression using client's data on laboratory analysis of breast
cancer tumours.
Automated Evaluation of Predictive Models
R code to compare the predictions from computational chemistry models
to observed values and generate a Word report and associated data
files.
Training
I have run general R courses in collaboration
with Prism,
based in Cambridge,
including Getting
Started with R and R
Programming. If you are interested in attending such a course, or
having a custom course run on-site, you are welcome to contact myself or
Prism.
