Jan. 25, 2012
Our gene sets based network inference papers published in Bioinformatics and TCBB.

Dec. 22, 2011

Dr. Lipi Acharya graduated with a PhD and landed her first job at Dow AgroSciences.

Sept. 16, 2011
Computer lab is being set up on the thrid floor of Maccabees Building.

Aug. 18, 2011
We have moved to Wayne Syate University Computer Science Department.


  • Reverse engineering biological networks:
We study the fundemental aspects of network structures and biological functions. One of the key problems that we are attempting to solve is the reverse engineering problem:  inferring the the latent network structures that emit the observed data and execute the anticipated biological functions. To this end we have developed a set novel stochastic models and algorithms. These algorithms perfom favorablly to the existing appraoches such as Bayesian Networks, Graphical Gaussian Models and Relevance Netowrks. On the biological application side, we are interested in the relationships between network structures and biological functions.       
Gene Set Gibbs Sampler
  • High throughput data analysis
One of the newer type of high thoughput molecular pofiling data is the next-generation sequencing data. This type of data is featured by ultra-high throughput (billions), sequencing errors and sampling bias. Currently we primarily focus on develping models, algorithms and tools for analyzing transcriptome sequencing data, i.e. RNA-seq. Som of more important problems we are attempting to address are: transcriptome de novo reconstruction and transcripts quantification. The former probem is complicated by the highly nonlinear transcript structures and ultra-high throughput of reads. The latter is chanllenged by the sequencing bias and errors. 
  • Open source GUI software development
User Interface is very important, if not more than, the software tools theselves. We work hard to make our software easily approachable by the end users. We use some of more advanced Java GUI techniques, such as JFace, to provide Graphical User Interface (GUI) for our software tools.

  • Pattern recognition, machine learning and data mining
At the moment we are primrily interested in multivariate statistical models and iterative algorithms for unsupervised learning from replicated and incomplete data. As the molecular profiling data is often noisy and incomplete it is necessary to use multivariate models to account for individual replicate and iterative algorithms to learn model parameters using incomplete data. We also incoporate experimental designs into our modelling. We develop such models and algorithms to learn complicated dependency structures from replicated and incomplete molecular profiling data.

Another related project is to developing analytical and numerical algorithms to solve discrete optimization problems in bioinformatics. One of our most recent research is to develop a Markov Chain Monte Carlo strategy to infer the gene regulatory networks.