 |
| The Lane Center for Computational Biology at Carnegie Mellon University seeks to realize the potential of machine learning for expanding our understanding of complex biological systems. A primary goal of the center is to develop computational tools that will enable automated creation of detailed, predictive models of biological processes, including automated experiment design and data acquisition. We anticipate that these efforts will not only lead to deep biological knowledge but also to tools for individualized diagnosis and treatment of disease. The Lane Center builds on the strong history of computational and interdisciplinary research at Carnegie Mellon. |
Voting Faculty
|
Affiliated Faculty |
Adjunct Faculty
|
Dr. Peter Berget's group is customizing single-chain antibody fragments (scFvs) that react with small molecule haptens. Through this "protein engineering" we are exploring the use of these scFvs as "biosensors" for specific reactions and pathways in mammalian cells. My laboratory uses molecular biology techniques to engineer single chain antibody fragments to react to specific intracellular signals, thus making them into "biosensors" for networks and pathways. We also using these techniques to design and construct protein tagging retroviral vectors. We are also collaborating with Bob Murphy's laboratory to develop high throughput techniques to analyze the location and dynamics of GFP tagged proteins in CD-tagged mammalian cell populations. My laboratory is developing high throughput methods to both tag cells and identify the tagged genes.
Dr. William Cohen is interested in information integration and machine learning, particularly information extraction, text categorization and learning from large datasets. He has worked recently with Bob Murphy on machine learning approaches for extracting information from text and images in biomedical publications. A sample project is SLIF, a system that analyzes the text and images in online journal articles to find information about the subcellular localization of proteins (http://murphylab.web.cmu.edu/services/SLIF/).
Dr. Markus Deserno is interested in biophysical problems related to cell membranes and their interaction with proteins. Using analytical theory and coarse-grained molecular dynamics simulations he studies for instance the effect of protein binding on membrane curvature and the back-effect of membrane curvature on protein interactions. The
insights gained by this research are relevant for a variety of cell biological phenomena, such as protein aggregation, vesicle formation, or viral budding. A CMU-based web-page is currently under construction. At present, useful research material can be found at Dr. Deserno's web-page at the MPI-P
Dr. Dannie Durand's lab investigates the processes by which new genes arise through three types of activities: the development of computational methods for whole genome analysis, application of those methods to genomic data, and the interpretation of the results in light of contemporary gene and organismal function. Current research efforts in the laboratory include statistical tests for recognizing significant patterns in gene organization on chromsomes; the impact of large scale duplication on the evolution of insulin; homology identification for multi-domain protein families; tree-based methods for estimating gene duplication times; the role of duplication in pathway evolution.
Dr. Christos Faloutsos is interested in data mining for streams and graphs, fractals, database performance, and indexing for multimedia and bio-informatics data.
Dr. Geoff Gordon is interested in (among other things) statistical models of difficult data (e.g. images, sequences, video, text) and computational learning theory.
Dr. Veronica Hinman is interested in how developmental gene regulatory networks evolve and how such system-level processes relate to the maintenance or change of body type form.
Dr. Jonathan Jarvik works to advance CD-tagging - a method for genomic and proteomic discovery and analysis that entails inserting reporter-encoding guest exons into genomic DNA. The CD-tagging approach enables the discovery of previously unknown or poorly characterized proteins; observation and quantitation of the location, abundance and dynamics of the tagged proteins in living cells and tissues; purification of tagged genes, transcripts and proteins for biochemical and/or functional analysis; discovery and analysis of posttranslational modifications and protein-protein interactions; and the development of new cell based assays. We are collaborating on a location proteomics project with the Murphy and Berget groups aimed at imaging tens of thousands of new CD-tagged clones, identifying the genes that are tagged in each clone, and classifying the protein location patterns in the lines using automated image analysis methods developed by the Murphy group. Additional efforts are aimed at CD-tagging embryonic stem cells and developing new molecular tools that employ fluorogen-based biosensors as part of an NIH-roadmap Technology Center for Networks and Pathways headed by Alan Waggoner.
Dr. Jelena Kovacevic's research interests are in the area of signal analysis and processing. She is currently working on problems related to automated processing and analysis of large bioimage databases, including acquisition, segmentation, classification, and others. These problems use a variety of tools, including multiresolution tools, such as bases and frames. Her theoretical research interests include building new multiresolution tools as well as algebraic theory of signal processing.
Dr. Christopher Langmead's group uses a combination of machine learning and formal methods to model and reason about the behaviors of complex biological systems. We work closely with our collaborators in three application domains: (i) molecular dynamics, (ii) cancer dynamics, and (iii) the dynamics of acute illness.
Dr. Philip Leduc's research focuses on linking mechanics to biochemistry through exploring the science of molecular to cellular biomechanics through nano- and micro-technology, control theory approaches, and computational biology. The link between mechanics and biochemistry has been implicated in a myriad of scientific and medical problems, from orthopedics and cardiovascular medicine, to cell motility and division, to signal transduction and gene expression. Most of these studies have been focused on organ-level issues, yet cellular and molecular level research has become essential over the last decade in this field thanks to the revolutionary developments in genetics, molecular biology, microelectronics, and biotechnology. Developing molecular and cellular biomechanics with relation to biochemistry promises for a bright future with potential impacts on genomics, proteomics, tissue engineering, and medical diagnostics. By examining these issues in novel manners including utilizing nanotechnology, BioMEMS, and computational biology, he explores the linkages among these disciplines. Also, through focusing on nature inspired design principles at the molecular and cellular levels, novel approaches to technology development will be enabled.
Dr. Javier Lopez's group is interested in the mechanisms and regulation of pre-mRNA splicing in normal and disease states. Projects include analysis of mechanisms of alternative splicing and identification on a genome-wide scale of auxiliary sequence motifs that direct the accuracy and efficiency of constitutive and alternative splicing. Another project involves the genome-wide prediction and functional analysis of recursive splicing in large introns of Drosophila and mammals. We are also investigating the organization and function of splicing regulatory networks.
Dr. Robert Murphy's group does both experimental and computational cell biology, with a particular emphasis on developing fully-automated methods to understand the subcellular locations of proteins and how they change during development or disease (location proteomics). We use machine learning methods to compare, classify and cluster spatiotemporal patterns in microscope images and construct generative models directly from images to capture the essence of each subcellular pattern as well as the variation in pattern from cell to cell. The goal is to identify all "subcellular location families", how they change (especially during oncogenesis) and to provide generative models for each family that can be incorporated into systems biology simulations. We are particularly interested in active learning approaches to create closed loop systems of interpretation, modeling, experiment planning and automated data acquisition, enabling automated scientific discovery.
Dr. Gustavo Rohde's interests are image processing, modeling and numerical methods with applications to cell biology. He is interested in developing methods for automatic extraction, analysis, and interpretation of biological information from microscopy image data. In particular, he is interested in developing methods for characterizing the spatiotemporal properties related to structure and function of cells and tissues directly from images.
Dr. Roni Rosenfeld directs project GATTACA (http://www.cs.cmu.edu/%7Egattaca/), which builds statistical and computational models of viruses and other rapidly evolving pathogens. The models are used to predict the phenotype of newly emerging strains, to locate crucial mutations to confer these phenotypes, and to optimize and accelerate the biomedical discovery process. Project GATTACA also develops novel visualization and interactive exploration tools for very large alignments of biological sequences.
Dr. Gordon Rule's computational biology research program is directed at developing automated methods of NMR data processing that merge approaches from computer science with known structural features of proteins to produce robust algorithms for data analysis. Our current work is directed at expanding the capabilities of our public domain assignment software, MONTE.
Dr. Nick Sahinidis is interested in computational optimization and its applications in protein structural alignment, protein structure prediction, macromolecular structure determination via experimental techniques, metabolic network inference and directed improvement, drug design. Currently very interested in protein structural (3D) alignment.
Dr. Jeffrey Schneider is interested in active learning algorithms for efficiently controlling the experimental process during the creation and fitting of biological models. I believe appropriate selection algorithms will yield an order of magnitude improvement in the model discovery process. My previous efforts in this area have focused on in vivo CNS drug discovery. A broad perspective on my lab's activities can be found at the link below.
Dr. Russell Schwartz works broadly on models and simulations of biological systems. One major areas of focus is stochastic simulation methods for studying macromolecular assembly systems, where he develops algorithms for fast simulation and applies them to problems in the biophysics of complex self-assembly. His second major area of focus is developing models and algorithms for analyzing genetic variation data, where he is currently primarily interested in problems of phylogenetics and population structure.
Dr. Shlomo Ta'asan is interested in mathematical modeling of the immune system with emphasis on integrating real data in model construction and verification, mathematical models for early detection of disease based on multiple markers in serum with application to cancer and multiscale modeling of biological systems with data integration from genes to proteins to physiological parameters - a black-box approach using system identification methods.
Dr. Nathan Urban's lab is currently focusing on understanding the physiological mechanisms underlying the functional and computational properties of brain neuronal networks. In particular, we are interested in describing the detailed anatomical and physiological properties of cells and synapses, and then constructing models that provide insight into how these physiological properties give rise to the functional circuits that transform and store information in the brain. My goal is to use these models to get at the underlying computations that these physiological systems can be seen as implementing.
Dr. Eric Xing develops statistical models and machine learning algorithms for biological network inference and characterization, cis-regulatory module decoding, temporal/spatial gene expression analysis, regulatory evolution modeling, quantitative trait locus mapping, genome polymorphism patterning, and population genetic analysis. He is applying these quantitative approaches to investigate the mechanisms of cancer development and metazoan morphagenesis. He is also interested in developing statistical machine learning methodologies including graphical models, Bayesian approaches, inference algorithms, and learning theories for analyzing and mining high-dimensional, longitudinal, and relational data; and their applications in text/image mining, vision, and natural language processing.
Dr. Alan Waggoner's research has focused on development of fluorescence-based detection systems for biology and biotechnology. Waggoner is currently leading the Molecular Biosensor and Imaging Center into development of molecular biosensors for studying protein regulatory processes in living cells and tissues.
http://www.andrew.cmu.edu/user/murphy
http://www.cs.cmu.edu/~russells/
