eprintid: 30149
rev_number: 13
userid: 5901
dir: disk0/00/03/01/49
datestamp: 2016-11-22 17:39:16
lastmod: 2017-11-01 14:02:36
status_changed: 2016-11-22 17:39:16
type: conference_item
metadata_visibility: show
eprint_status: archive
creators_name: Burger, Virginia
creators_name: Bahar, Ivet
creators_name: Chennubhotla, Chakra
creators_email: 
creators_email: bahar@pitt.edu
creators_email: 
creators_id: 
creators_id: BAHAR
creators_id: 
title: A hierarchical elastic network model for unsupervised electron density map segmentation
ispublished: pub
divisions: sch_med_Computational_Systems_Biology
full_text_status: public
pres_type: poster
note: Competing Interests: No relevant conflicts of interest declared.
abstract: To highlight different levels of structural and functional organization found in cryo EM density maps, we use an unsupervised hierarchical elastic network model (hENM) based on a Markov diffusion process. The hENM segments electron density maps into meaningful protein subregions Chennubhotla & Bahar 2006, and finds a simplified representation of the map in each level of the hierarchy. When implemented on maps in the 2010 Cryo-EM Modeling Challenge, hENM segmented the molecules into individual proteins and regions without manual input. For example, hENM identified the apical and equatorial domains of the 14 groEL proteins in the 23.5A groEL-groES map (EMD: 1046). These results suggest that hENM could be applied to segment density maps which otherwise would require manual curating. As the sub-regions found in each level of the hierarchy highlight the core structural regions of the protein, they can also be used as anchor points for fitting high resolution structures into EM maps. Inference of secondary structure from characteristic patterns in the affinity maps can also help guide the fitting process. Further implementations of our method will study protein dynamics at the coarsest hierarchy level, and then project insight into dynamics onto finer, more detailed levels. This method provides a unifying framework in allowing protein segmentation concurrently with the study of information propagation through the network, with dynamical inference as an additional benefit.
date: 2011-02-07
date_type: published
publication: F1000 Research
publisher: F1000 Research Ltd.
event_title: Pacific Symposium on Biocomputing (PSB) 2011
event_type: conference
refereed: FALSE
issn: 2046-1402
official_url: https://f1000research.com/posters/807
citation:   Burger, Virginia and Bahar, Ivet and Chennubhotla, Chakra  (2011) A hierarchical elastic network model for unsupervised electron density map segmentation.  In: Pacific Symposium on Biocomputing (PSB) 2011.     
document_url: http://d-scholarship-dev.library.pitt.edu/30149/1/807.pdf
document_url: http://d-scholarship-dev.library.pitt.edu/30149/7/licence.txt