Sparse estimation for structural variability 英文参考文献.docVIP

Hosuretal.AlgorithmsforMolecularBiology2011,6:12 /content/6/1/12 RESEARCH OpenAccess Sparseestimationforstructuralvariability RaghavendraHosur1,3,RohitSingh1andBonnieBerger1,2* Abstract Background:Proteinsaredynamicmoleculesthatexhibitawiderangeofmotions;oftentheseconformational changesareimportantforproteinfunction.Determiningbiologicallyrelevantconformationalchanges,ortrue variability,efficientlyischallengingduetothenoisepresentinstructuredata. Results:Inthispaperwepresentanovelapproachtoelucidateconformationalvariabilityinstructuressolved usingX-raycrystallography.Wefirstinferanensembletorepresenttheexperimentaldataandthenformulatethe identificationoftrulyvariablemembersoftheensemble(asopposedtothosethatvaryonlyduetonoise)asa sparseestimationproblem.Ourresultsindicatethatthealgorithmisabletoaccuratelydistinguishgenuine conformationalchangesfromvariabilityduetonoise.WevalidateourpredictionsforstructuresintheProteinData BankbycomparingwithNMRexperiments,aswellasonsyntheticdata.Inadditiontoimprovedperformanceover existingmethods,thealgorithmisrobusttothelevelsofnoisepresentinrealdata.InthecaseofHuman Ubiquitin-conjugatingenzymeUbc9,variabilityidentifiedbythealgorithmcorrespondstofunctionallyimportant residuesimplicatedbymutagenesisexperiments.Ouralgorithmisalsogeneralenoughtobeintegratedintostate- of-the-artsoftwaretoolsforstructure-inference. Introduction when modeling NMR data. It has been suggested that, A central tenet of molecular biology is that a protein’s foranaccuraterepresentationofthephysicalheteroge- three-dimensional (3D) structure is crucial to its func- neityinaprotein,suchmultiple-conformermodelsalso tion.Indeedthestructuralgenomicsinitiativeisprodu- beusedtoexplainX-raycrystallographydata[8-10]. cing an ever increasing number of structures at high An open problem– and the focus of this paper– is resolution,providingaccuratecoordinatesforeachatom understanding the nature of conformational variability in the structure [1]. A protein’s s