Software, Algorithms and Patents

  • ProGENI (Prioritization of Genes Enhanced with Network Information)

ProGENI is a computational method to identify genes whose basal mRNA expression can predict the sensitivity of tumor cells to different treatments by leveraging prior knowledge in the form of protein-protein and genetic interactions. ProGENI is based on identifying a small set of genes where a combination of their expression and the activity level of the network module surrounding them shows a high correlation with drug response, followed by the ranking of the genes based on their relevance to this set using random walk techniques.

Code of the algorithm.

A talk on KnowEnG and ProGENI (Cold Spring Harbor Laboratory 2016).

A. Emad, J. Cairns, K. R. Kalari, L. Wang, S. Sinha, “Knowledge-guided gene prioritization reveals new insights into the mechanisms of chemoresistance,” Genome Biology, 18(1), 153, 2017

 

  • C3 (Cancer Correlation Clustering)

C3 is a method based on community detection techniques to analyze combinatorial patterns of cancer alterations. This method, leverages mutual exclusivity of mutations, patient coverage and driver network concentration principles to identify mutation patterns and driver pathways in various types of cancer.

Code of the algorithm.

J. P. Hou*, A. Emad*, G. J. Puleo, J. Ma, and O. Milenkovic, “A new correlation clustering method for cancer mutation analysis,” Bioinformatics, 32 (24), pp. 3717-3728, 2016. *Contributed equally

 

  • Methods and systems for determining crosstalk for a line in a vectored system

A. Emad, C. J. Nuzman, and E. Soljanin. “Methods and systems for determining crosstalk for a line in a vectored system.” U.S. Patent Application No. 14/669,167, filed March 26, 2015.

 

  • CaSPIAN (Causal Subspace Pursuit for Inference and Analysis of Networks)

CaSPIAN is a novel algorithm for inference of directed edges in a network based on compressive sensing and Granger causality. This algorithm was specifically developed for inference of causal interactions in gene regulatory networks, but can be used for any general network containing directed edges.

Code of the algorithm.

A. Emad and O. Milenkovic, “CaSPIAN: A Causal Compressive Sensing Algorithm for Discovering Directed Interactions in Gene Networks,” PLOS ONE, vol. 9, no. 3, e90781, March 2014. doi:10.1371/journal.pone.0090781.

 

  • RFIT (Residual Feedback Iterative Thresholding)

RFIT is a novel algorithm for solving the low-rank matrix recovery problem. This algorithm introduces a novel choice of update direction inspired by the
Approximate Message Passing (AMP) compressed sensing solver and Heavy Ball method in optimization theory.

 Code of the algorithm: please email me.

P. Johnstone, A. Emad, O. Milenkovic, and P. Moulin, “RFIT: A New Algorithm for Matrix Rank Minimization,” Signal Processing with Adaptive Sparse Structured Representations (SPARS’13), 2013.

 

  • Message Passing Decoder for Semi-Quantitative Group Testing Framework

Semi-quantitative group testing (SQGT) is a novel group testing method motivated by a class of problems arising in genome screening experiments. SQGT is a
(possibly) non-binary pooling scheme that may be viewed as a concatenation of an adder channel and an integer-valued quantizer. In its full generality, SQGT can be viewed as a unifying framework for group testing, in the sense that most group testing models are special instances of SQGT. We have described different code constructions for SQGT in [Emad and Milenkovic (2012)]. In addition, different decoders are introduced including a belief propagation decoder for sparse SQGT codes.

 Code of the algorithm: please email me.

A. Emad and O. Milenkovic, “Semiquantitative Group Testing,” IEEE Trans. Inf. Theory, vol. 60, pp. 4614-4636, 2014