Regulatory Networks

Application of the Dynamic Regulatory Events Miner (DREM) to study amino acid starvation response in yeast.


Our group develops computational methods for understanding the dynamics, interactions and conservation of complex biological systems. As new high-throughput biological data sources become available, they hold the promise of revolutionizing molecular biology by providing a large-scale view of cellular activity. However, each type of data is noisy, contains many missing values and only measures a single aspect of cellular activity. Our computational focus is on methods for large scale data integration. We primarily rely on machine learning and statistical methods. Most of our work is carried out in close collaboration with experimentalists. Many of the computational tools we develop are available and widely used.

Read more

Latest Publications

Full list of publications


  • 12/2019

    Deep learning for gene interaction prediction paper published in PNAS We presented a new method for encoding gene expression data as an image like input. As we showed, using such encoding we can utilize the power of convolutional neural networks for several tasks related to inferring gene interactions. The new method, termed CNNC, has already been applied to identify new partners for genes, reconstruct pathways based on these interactions and assign new function to genes. In all cases the new encoding helps the method outperform all prior methods for these important tasks. You can read the full paper in PNAS and view the press release issued by CMU to discuss the importance of the work.

  • 11/2019

    Sabrina's thesis awarded SCS Honorable Mention Sabrina Rashid's PhD thesis, 'Distributed Computing in Nature', is one of 5 from the School of Computer Science at CMU selected for a PhD thesis award. Given the large number of PhD students at SCS and the amazing work they all do, this is a great honor! Congratulations to Sabrina.

  • 10/2019

    HuBMAP paper published in Nature The HuBMAP perspective paper, is published in Nature. The paper presents the vision and goals and the HuBMAP consortium. We are leading a large group of researchers at Carnegie Mellon, the University of California Santa Cruz and the Sanger Institute in the UK which is in charge of the computational tools and pipeline development for HuBMAP. See the SCS press release which includes a quote from Matt Ruffalo who is leading the development team.

Read more