We have published a pre-print manuscript for our clust method, which we believe to solve the biggest problem in gene expression clustering, from our point of view:
The manuscript shows how clust solves the problem by application to 100 different microarray and RNAseq gene expression datasets from 5 model species. Clust is thoroughly compared with seven widely used methods (k-means, MCL, hierarchical clustering, WGCNA, self-organising maps (SOMs), Click, and Cross-Clustering).
Update: the paper has been accepted for publication in Genome Biology and will be available online soon!
After about three and a half hours of discussion of my thesis, the chair of the examination, Professor John Stonham, informed me that I qualify to be awarded the Doctor of Philosophy (PhD) degree conditional upon applying minor corrections to my thesis. Corrections are planned to be submitted within few weeks.
The exam was conducted by the external examiner, Professor Mahesan Niranjan from the Department Electronics and Computer Science (ECS) at the University of Southampton, and the internal examiner, Dr Ali Mousavi, from the Department of Electronic and Computer Engineering (ECE) at Brunel University London.
I found the discussion very interesting, enjoyable, thorough, and informative. It allowed me to reinforce my confidence in the quality of my research while pointing out few aspects which need enhancing. For example, the choice of the individual clustering methods that are used at to feed the Bi-CoPaM and the UNCLES methods should be reconsidered.
Some other ideas were suggested by the examiners which are worth trying! I would like to thank both of them as well as the chair of the exam for the very interesting and engaging discussion. I really didn’t feel that they were 3.5 hours 🙂