CyclinPred is a SVM based prediction
method to identify novel cyclins using various
features of proteins.
Cyclins are a class
of eukaryotic proteins which are crucial in cell-cycle control of an organism, first identified in early
1980s that are characterized by their periodic
accumulation during interphase followed by rapid
degradation at each meiotic or mitotic division.
The diverse nature of cyclins in
terms of sequence as well as its functionality made
its prediction as an important and challenging problem
in the biological world. The Prediction of Novel
Cyclins can provide deep insight into new key players of
cell-cycle and its various functional aspects as a
function of different life-cycles in different
organisms. Hence, a new prediction protocol needed to
be designed and evaluated which is independent of
sequential or structural homology to proteins with
known functions. Here, the features used in CyclinPred are: amino acid
composition, dipeptide composition, secondary structure composition and PSSM composition profile of proteins were
considered for the development of method. Hybrid modules were also designed using these protein features.
Depending on comparative evaluation of all models, blind-test performance and various statistical measures, the best model was decided.
Here, we were able to achieve the prediction accuracy of 92.14% for PSSM based model.
The current SVM model can predict both groups of cyclins: G1/S cyclins and G2/M cyclins.
Related Publication: PLoS ONE. 2008; 3(7): e2605