Brief Description of projects

1) Projects funded by large research grants



Solar bio-fuel and carbon sequestration with Cyanobacteria: role of genetic network

Funded by Australia-India Strategic Research Fund (AISRF) initiative (Combined funding over A$625 k): The aim is to develop computer based models using microarray data for four different cyanobacteria strains. With large scale implementation, obtain network motifs responsible for carbon sequestration and/or solar bio-fuel generation.

Team members:

 Research Fellow based at Monash University: Dr. Vinh Nguyen   

Senior Research Fellow at IIT Bombay: Dr Sandeep Gaudana 

Research Associate at IIT Bombay: Ms Madhuri Digmurthi

Chief Investigators : A/Prof Madhu Chetty and Prof Ross Coppel (Monash) and Prof Pramod Wangikar (Indian Institute of Technology, Bombay)




Evolutionary search techniques for protein folding prediction.

Funded by Australian Research Council (ARC) Discovery Project (A$200K): Aim is to develop accurate protein models and evolutionary search techniques which can be implemented on computational grid infrastructure.

Team members: Tamjidul Hoque (Research Fellow), Chief Investigators: Abdul Sattar (Griffith),  Andrew Lewis (Griffith) and  Madhu Chetty (Monash) 


2) On-going doctoral projects


 To be finalised (Commencing May 2014)

Team members: Rubaiya Rahtin Khan (doctoral student- yet to commence), Madhu Chetty (Main)


State Space Approach for modelling genetic network

aim is to apply well known state space approach for modelling genetic network interactions. We intend to investigate low order model to capture genetic interactions to overcome the "curse of dimensionality". There is very limited work done in this area. It is proposed to explore linear dynamic model techniques from the field of system theory to model genetic interactions.

Team members: Ahammed K Yousef (Doctoral student- Current), Madhu Chetty (Main Supervisor), Gour Karmakar (Associate), Terry Caelli from NICTA (External Supervisor)



Maximal H-Core formation for low resolution protein structure prediction

aim is to develop techniques based on maximal H-Core for low resolution HP protein models. While H-Core approach is deterministic, we propose to implement this within a evolutionary framework improving upon the factors such as population diversity to provide a quick and accurate search in the complex landscape for protein folding prediction. The techniques will be applied for sequences with large lengths.

Team members: Rumana Nazmul (Doctoral student - Current), Madhu Chetty (Main Supervisor), Ram Samudrala from University of Washington (External), David Chalmers from Pharmacy faculty of Monash(Associate)



 Incorporating a-priori knowledge in dynamic network modelling of genetic networks

modelling of genetic network using Bayesian approach suffers from very few samples and very high number of variables. To overcome the limitations due to this curse of dimensionality, we propose to develop methodology of including a-priori knowledge (in form of knowledge of genetic interactions and also knowledge of transcription factors) to make the modelling process accurate and reliable. Further, novel BN learning techniques will be developed to work with incorporation with incorporation of prior knowledge. The work will be essentially focussed on modelling cyanobacterial strain Cyanothece 51142 for understanding its role for carbon sequestration and solar bio-fuel.

Team members: Ajay Nair (Doctoral student- under Monash/IITB academy), Madhu Chetty (Joint Supervisor), Pramod Wangikar (Joint Supervisor)



Enhanced and desirable lipid synthesis in algae for efficient biofuel production

Use of microalgae considered as a suitable alternative feedstock for   biofuels because certain species contain high amounts of lipids, which could be extracted, processed and refined into suitable fuels. The aim is to understand lipid biosynthetic pathways in microalgae, identifying key regulatory steps, designing micro RNAs for specific pathways which enables desired lipid and test their validity. The outcome of this project would be algal strains with specific microRNA mode of action enable to produce around 25-30% lipid on dry weight basis.

Team members: Akhila George (Doctoral student- under Monash/IITB academy), John Beardall and Madhu Chetty (Joint Supervisor), Pramod Wangikar, IITB (Joint Supervisor)



3) Completed doctoral projects


Dynamic Bayesian network for modeling of genetic networks

aim to model and infer genetic networks with dynamic Bayesian network (DBN) and the non-linear mutual information as the scoring metric - so as to penalise complexities in the network. Applying DBN, we aim to also be able to identify the feedback loops and the time delays in the genetic interactions. Synthetic data and real life yeast cell cycle data sets will be used. The model will be extended to study cyanobacteria for solar bio-fuel.

Key journal publication: BMC Systems Biology

Team members: Nizamul Morshed (Doctoral student - Completed), Madhu Chetty (Main Supervisor), Terry Caelli from NICTA (External)


S-System approach for modeling of genetic networks

model for describing biochemical networks is rich enoughto reasonably capture the nonlinearity of genetic regulation. S-system model is based on a set of non-linear ordinary differential equation in which the component processes are characterized by power-law functions. We aim to develop novel evolutionary techniques to determine the S-system model parameters. Also, rather than getting restricted to simple toy problems containing few genes, we aim to investigate large scale realistic systems with the aid of computational grid. The model will be extended to investigate cyanobacteria for carbon sequestration.

Team members: Ahsan Raja Chowdhury (Doctoral student -thesis submitted), Madhu Chetty (Main Supervisor)

 Key journal publication: BMC Bioinformatics, Elsevier's Cognitive Neurodynamics


Feature selection in micro array gene expression data

The project developed filter-based feature selection techniques suitable for datasets with high dimensionality and large number of classes or more specifically, multiclass gene expression datasets such as the Global Cancer Map (GCM) and the NCI60 datasets.  The techniques were able to produce feature subsets or predictor sets capable of giving suitably high classification rates, while using a minimal number of features.

key journal publications: 1) BMC Bioinformatics 2) Data mining and knowledge discovery 3) Algorithms for molecular biology

Key journal publications: 1) Data Mining and Knowledge Discovery 2) BMC Bioinformatics 3) Algorithms for Molecular Biology

Members: Chia Huey Ooi (Doctoral student -Completed), Madhu Chetty (Main Supervisor),  Shyh Wei Teng (Associate)



Hybrid computational models for protein sequence analysis and secondary structure prediction

pIn this project, we developed novel hybrid computational models for protein sequence analysis and secondary structure prediction. These models will be used to explore profiles for homologous proteins, structure prediction, missing data estimation and inferring of phylogenetic trees.

Key journal publications in: 1) Elsevier's Neurocomputing 2) Journal of intelligent and fuzzy systems 3) Elsevier's journal of genomics, proteomics and bioinformatics

Team Members: Niranjan Bidargaddi (Doctoral student- Completed),  Madhu Chetty (Main Supervisor),  J. Kamruzzaman (Associate)



Genetic Algorithms for Ab Initio Protein folding prediction using low resolution HP models

       We developed theoretical underpinnings and a systematic methodology for HP Model for improving the accuracy including fitness evolution and computational throughput in Protein Folding Prediction (PFP),  extended these strategies to 3D HP and FCC models to predict the ab initio three dimensional structure of a protein from its primary amino acid sequence. Further, we also proposed strategies for handling large amino acid sequences.

Key journal publications: 1) IEEE Trans in Computational Biology and Bioinformatics 2) Journal of Computational Biology

3) Elsevier's Neurocomputing journal

Team Members: Tamjidul Hoque (Doctoral student- Completed), Madhu Chetty (Main Supervisor), Abdul Sattar from Griffith Uni (External)



Markov Blanket Based Causal Modeling and Inferencing of Gene regulatory network

aim of the project are to primarily look into the modeling and inferencing of GRNs based on Causal models and Genetic Algorithm, and their subsequent implementation and testing using real temporal data sets (such as yeast), available in public domain. Understanding of GRNs is still at a relatively embryonic stage and is the next big break for computational biology. It is very difficult to understand biological systems because of their inherent complexities. The main significance of this project will be the development of novel algorithms and techniques for modeling that will help in successful deduction of the said functions for each gene node, which in turn will lead to more accurate modeling behaviour of a cell.

Key journal publication: 1) IEEE Trans in Computational Biology and Bioinformatics

 Team Members: Ramesh Ram (Doctoral student-Completed),  Madhu Chetty (Main Supervisor) and Dieter Bulach, CSIRO (External)



Memetic algorithm for protein structure prediction using low resolution lattice models

aim is to improve upon both aspects -global search and local search to provide a quick and accurate search in the complex landscape for protein folding prediction. For this, amongst other things, we propose to investigate systematic population generation eliminating SAW (self avoiding walks), clustering of population in different basins of attractions to reach global optima. We also propose a new concept of meme based generation of population for fast exploration of search space.

Key journal publication: 1) IEEE Trans in Evolutionary Computation

Team members: Kamrul Islam (Doctoral Student -Completed, Madhu Chetty (Main Supervisor), Manzur Murshed (Associate)