Applying systems approaches to biofuel research: prediction of function of plant polysaccharide synthesis enzymes

Paul Dupree & Marcelo Segura

Questions and issues that arise include:
a) How can we objectively judge and report the quality of functional protein clustering
in LOPIT and ProCoDeS datasets using different clustering methods?
b) How can we merge the data from multiple replicate experiments to strengthen predictions, rather than adding noise? Since each experiment has subtly different sampling parameters, the data cannot be averaged. Moreover, unlike transcriptomic data, proteomic data misses observation of a substantial number of proteins. Therefore,
each protein may not be observed in each experiment.
c) How can we best integrate the different types of data to make stronger
predictions?

Proceedings of the 2nd Mathematics in the Plant Sciences Study Group