Evaluation of methods for classification using transcriptome data
In work carried out by two undergraduate students doing their final year project at NCL, we evaluated the performance of thousands of combinations of algorithms used to select features from high-dimensional DNA microarray data, and use the values of individual observations for those features to classify the data into different cancer types. Details can be found in a paper that was published in 2008.
Students: Vinaya Vijayan, Nadeem Bulsara
Collaborator: Mugdha Gadgil, NCL
Evaluation of methods for identifying periodically-expressed genes
We developed a new method for identifying truly periodic genes from time-course transcriptomic data and identified a new way of objectively comparing the performance of such algorithms. A preprint of the article is available on request.
Students: Vinaya Vijayan and Prachi Deshpande
Â
Collaborator: Mugdha Gadgil, NCL