{"657150":{"#nid":"657150","#data":{"type":"news","title":"AF2Complex: Researchers Leverage Deep Learning to Predict Physical Interactions of Protein Complexes","body":[{"value":"\u003Cp\u003EFrom the muscle fibers that move us to the enzymes that replicate our DNA, proteins are the molecular machinery that makes life possible.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EProtein function heavily depends on their three-dimensional structure, and researchers around the world have long endeavored to answer a seemingly simple inquiry to bridge function and form:\u0026nbsp;if you know the building blocks of these molecular machines, can you predict how they are assembled into their functional shape?\u003C\/p\u003E\r\n\r\n\u003Cp\u003EThis question is not so easy to answer. With complex structures dependent on intricate physical interactions, researchers have turned to artificial neural network models \u0026ndash; mathematical frameworks that convert complex patterns into numerical representations \u0026ndash; to predict and \u0026ldquo;see\u0026rdquo; the shape of proteins in 3D.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EIn a new\u0026nbsp;\u003Ca href=\u0022https:\/\/www.nature.com\/articles\/s41467-022-29394-2\u0022 target=\u0022_blank\u0022\u003Epaper\u003C\/a\u003E\u0026nbsp;published in\u0026nbsp;\u003Cem\u003ENature Communications\u003C\/em\u003E, researchers at Georgia Tech and\u0026nbsp;\u003Ca href=\u0022https:\/\/www.ornl.gov\/\u0022 target=\u0022_blank\u0022\u003EOak Ridge National Laboratory\u003C\/a\u003E\u0026nbsp;build upon one such model,\u0026nbsp;\u003Ca href=\u0022https:\/\/www.deepmind.com\/research\/highlighted-research\/alphafold\u0022 target=\u0022_blank\u0022\u003EAlphaFold 2\u003C\/a\u003E, to not only predict the biologically active conformation of individual proteins, but also of functional protein pairings known as complexes.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EThe work could help researchers bypass lengthy experiments to study the structure and interactions of protein complexes on a large scale, said\u0026nbsp;\u003Ca href=\u0022https:\/\/biosciences.gatech.edu\/people\/jeffrey-skolnick\u0022 target=\u0022_blank\u0022\u003EJeffrey Skolnick\u003C\/a\u003E, Regents\u0026rsquo; Professor and Mary and Maisie Gibson Chair in the\u0026nbsp;\u003Ca href=\u0022https:\/\/biosciences.gatech.edu\/\u0022 target=\u0022_blank\u0022\u003ESchool of Biological Sciences\u003C\/a\u003E\u0026nbsp;and one of the corresponding authors of the study, adding that computational models such as these could mean big things for the field.\u0026nbsp;\u003C\/p\u003E\r\n\r\n\u003Cp\u003EIf these new computational models are successful, Skolnick said, \u0026ldquo;it could fundamentally change the way biological molecular systems are studied.\u0026rdquo;\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003EPrimed for Protein Prediction\u003C\/strong\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003ECreated by London-based artificial intelligence lab\u0026nbsp;\u003Ca href=\u0022https:\/\/www.deepmind.com\/\u0022 target=\u0022_blank\u0022\u003EDeepMind\u003C\/a\u003E, AlphaFold 2 is a deep learning neural network model designed to predict the three-dimensional structure of a single protein given its amino acid sequence. Skolnick and fellow corresponding author,\u0026nbsp;\u003Ca href=\u0022https:\/\/biosciences.gatech.edu\/people\/mu_gao\u0022 target=\u0022_blank\u0022\u003EMu Gao\u003C\/a\u003E, senior research scientist in the School of Biological Sciences, shared that the Alphafold 2 program was\u0026nbsp;\u003Ca href=\u0022https:\/\/www.nature.com\/articles\/d41586-020-03348-4\u0022 target=\u0022_blank\u0022\u003Ehighly successful\u003C\/a\u003E\u0026nbsp;in blind tests occurring at the 14\u003Csup\u003Eth\u0026nbsp;\u003C\/sup\u003Eiteration of the Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction, or\u0026nbsp;\u003Ca href=\u0022https:\/\/predictioncenter.org\/casp14\/index.cgi\u0022 target=\u0022_blank\u0022\u003ECASP14\u003C\/a\u003E, a bi-annual competition where researchers around the globe gather to put their computational models to the test.\u0026nbsp;\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u0026ldquo;To us, what is striking about AlphaFold 2 is that it not only makes excellent predictions on individual protein domains (the basic structural or functional modules of a protein sequence), but it also performs very well on protein sequences composed of multiple domains,\u0026rdquo; Skolnick shared. And so with the ability to predict the structure of these complicated, multi-domain proteins, the research team set out to determine if the program could go a little further.\u0026nbsp;\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u0026ldquo;The physical interactions between different [protein] domains of the same sequence are essentially the same as the interactions gluing different proteins together,\u0026rdquo; Gao explained.\u0026nbsp;\u0026ldquo;It quickly became clear that relatively simple modifications to AlphaFold 2 could allow it predict the structural models of a protein complex.\u0026rdquo; To explore different strategies,\u0026nbsp;\u003Ca href=\u0022https:\/\/davinan.github.io\/dna\/\u0022 target=\u0022_blank\u0022\u003EDavi Nakajima An\u003C\/a\u003E, a fourth-year undergraduate in the School of Computer Science, was recruited to join the team\u0026rsquo;s effort.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EInstead of plugging in the features of just one protein sequence into AlphaFold 2 per its original design, the researchers joined the input features of multiple protein sequences together. Combined with new metrics to evaluate the strength of interactions among probed proteins, their new program\u0026nbsp;\u003Ca href=\u0022https:\/\/github.com\/FreshAirTonight\/af2complex\u0022 target=\u0022_blank\u0022\u003EAF2Complex\u003C\/a\u003E\u0026nbsp;was created.\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003ECharting New Territory\u003C\/strong\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003ETo put AF2Complex to the test, the researchers\u0026nbsp;partnered with the high-performance computing center,\u0026nbsp;\u003Ca href=\u0022https:\/\/pace.gatech.edu\/\u0022 target=\u0022_blank\u0022\u003EPartnership for an Advanced Computing Environment\u003C\/a\u003E\u0026nbsp;(PACE), at Georgia Tech,\u0026nbsp;and charged the model with predicting the structures of protein complexes it had never seen before. The modified program was able to correctly predict the structure of over twice as many protein complexes as a more traditional method called docking. While AF2Complex only needs protein sequences as input, docking relies on knowing individual protein structures beforehand to predict their combined structure based on complementary shapes.\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u0026ldquo;Encouraged by these promising results, we extended this idea to an even bigger problem, which is to predict interactions among multiple arbitrarily chosen proteins, e.g., in a simple case, two\u0026nbsp;arbitrary proteins,\u0026rdquo; shared Skolnick.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EIn addition to predicting the structure of protein complexes, AF2Complex was charged with identifying which of over 500 pairs of proteins were able to form a complex at all. Using newly designed metrics, AF2Complex outperformed conventional docking methods and AlphaFold 2 in identifying which of the arbitrary pairs were known to experimentally interact.\u003C\/p\u003E\r\n\r\n\u003Cp\u003ETo test AF2Complex on the proteome scale, which encompasses an organism\u0026rsquo;s entire library of the proteins that can be expressed, the researchers turned to the\u0026nbsp;\u003Ca href=\u0022https:\/\/www.olcf.ornl.gov\/summit\/\u0022 target=\u0022_blank\u0022\u003ESummit Oak Ridge Leadership Computing Facility\u003C\/a\u003E, the world\u0026rsquo;s second largest supercomputing center. \u0026ldquo;Thanks to this resource, we were able to apply AF2Complex to about 7,000 pairs of proteins from the bacteria\u0026nbsp;\u003Cem\u003EE. coli\u003C\/em\u003E,\u0026rdquo; Gao shared.\u0026nbsp;\u003C\/p\u003E\r\n\r\n\u003Cp\u003EIn that test, the team\u0026rsquo;s new model not only identified many pairs of proteins known to form complexes, but it was able to provide insights into interactions \u0026ldquo;suspected but never observed experimentally,\u0026rdquo; Gao said.\u0026nbsp;\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDigging deeper into these interactions revealed a potential molecular mechanism for protein complexes\u0026nbsp;that are\u0026nbsp;particularly important for energy transport. These protein complexes are known to carry hemes, essential metabolites giving blood dark red color. Using AF2Complex\u0026rsquo;s predicted structural models, \u003Ca href=\u0022https:\/\/cmb.ornl.gov\/jerry-m-parks\/\u0022 target=\u0022_blank\u0022\u003EJerry M. Parks\u003C\/a\u003E, a senior research and development staff scientist at Oak Ridge National Laboratory and a collaborator in the study, was able to place hemes at their suspected reaction sites within the structure. \u0026ldquo;These computational models now provide insights into the molecular mechanisms for how this biomolecular system works,\u0026rdquo; Gao said.\u0026nbsp;\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u0026ldquo;Deep learning is changing the way one studies a biological system,\u0026rdquo; Skolnick added. \u0026ldquo;We envision methods like AF2Complex will become powerful tools for any biologist who would like to understand molecular mechanisms of a biosystem involving protein interactions.\u0026rdquo;\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003EAF2Complex is an open-source tool available to the public and can be downloaded\u0026nbsp;\u003C\/strong\u003E\u003Ca href=\u0022https:\/\/github.com\/FreshAirTonight\/af2complex\u0022 target=\u0022_blank\u0022\u003E\u003Cstrong\u003Ehere\u003C\/strong\u003E\u003C\/a\u003E\u003Cstrong\u003E.\u003C\/strong\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cem\u003EThis work was supported in part by the DOE Office of Science, Office of Biological and Environmental Research (DOE DE-SC0021303) and the Division of General Medical Sciences of the National Institute Health (NIH R35GM118039). DOI:\u0026nbsp;\u003C\/em\u003E\u003Ca href=\u0022https:\/\/doi.org\/10.1038\/s41467-022-29394-2\u0022 target=\u0022_blank\u0022\u003E\u003Cem\u003Ehttps:\/\/doi.org\/10.1038\/s41467-022-29394-2\u003C\/em\u003E\u003C\/a\u003E\u003C\/p\u003E\r\n","summary":null,"format":"limited_html"}],"field_subtitle":[{"value":"A computational tool developed to predict the structure of protein complexes \u2013 the molecular machinery that makes life possible \u2013 is lending new insights into the biomolecular mechanisms of their function. "}],"field_summary":[{"value":"\u003Cp\u003EProteins are the molecular machinery that makes life possible, and researchers have long been interested in a key trait of protein function: their three-dimensional structure.\u0026nbsp;\u0026nbsp;A new study by Georgia Tech and Oak Ridge National Laboratory details a computational tool able to predict the structure of protein complexes \u0026ndash; and lends new insights into the biomolecular mechanisms of their function.\u003C\/p\u003E\r\n","format":"limited_html"}],"field_summary_sentence":[{"value":"A computational tool developed to predict the structure of protein complexes \u2013 the molecular machinery that makes life possible \u2013 is lending new insights into the biomolecular mechanisms of their function. "}],"uid":"35575","created_gmt":"2022-04-11 14:42:23","changed_gmt":"2022-04-18 13:10:52","author":"adavidson38","boilerplate_text":"","field_publication":"","field_article_url":"","dateline":{"date":"2022-04-18T00:00:00-04:00","iso_date":"2022-04-18T00:00:00-04:00","tz":"America\/New_York"},"extras":[],"hg_media":{"657354":{"id":"657354","type":"image","title":"Researchers Jeffrey Skolnick and Mu Gao at the Engineered Biosystems Building at Georgia Tech. (Photo: Jess Hunt-Ralston)","body":null,"created":"1650045007","gmt_created":"2022-04-15 17:50:07","changed":"1650045007","gmt_changed":"2022-04-15 17:50:07","alt":"","file":{"fid":"249155","name":"2022 04 Jeffrey Skolnick and Mu Gao - Biosci research copy.jpg","image_path":"\/sites\/default\/files\/images\/2022%2004%20Jeffrey%20Skolnick%20and%20Mu%20Gao%20-%20Biosci%20research%20copy.jpg","image_full_path":"http:\/\/tlwarc.hg.gatech.edu\/\/sites\/default\/files\/images\/2022%2004%20Jeffrey%20Skolnick%20and%20Mu%20Gao%20-%20Biosci%20research%20copy.jpg","mime":"image\/jpeg","size":2689047,"path_740":"http:\/\/tlwarc.hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/images\/2022%2004%20Jeffrey%20Skolnick%20and%20Mu%20Gao%20-%20Biosci%20research%20copy.jpg?itok=LXto2eZ_"}},"657142":{"id":"657142","type":"image","title":"A 3D rendering of a protein complex structures predicted from protein sequences by AF2Complex. Credit: Mu Gao.","body":null,"created":"1649684817","gmt_created":"2022-04-11 13:46:57","changed":"1649684817","gmt_changed":"2022-04-11 13:46:57","alt":"A 3D rendering of the structures of three\u00a0protein complexes, predicted from protein sequences by AF2Complex.","file":{"fid":"249063","name":"AF2Complex_PredictedProteins-01.png","image_path":"\/sites\/default\/files\/images\/AF2Complex_PredictedProteins-01.png","image_full_path":"http:\/\/tlwarc.hg.gatech.edu\/\/sites\/default\/files\/images\/AF2Complex_PredictedProteins-01.png","mime":"image\/png","size":3339555,"path_740":"http:\/\/tlwarc.hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/images\/AF2Complex_PredictedProteins-01.png?itok=oLd6_9df"}},"657144":{"id":"657144","type":"image","title":"The initial development of AF2Complex was done at the Partnership for an Advanced Computing Environment (PACE) computing center of Georgia Tech, pictured here in the Coda Data Center.\u00a0Credit: Paul Manno\/PACE.","body":null,"created":"1649685349","gmt_created":"2022-04-11 13:55:49","changed":"1649685349","gmt_changed":"2022-04-11 13:55:49","alt":"A row of computer servers illuminated with blue light in a white hallway.","file":{"fid":"249065","name":"hive_img_8791.jpg","image_path":"\/sites\/default\/files\/images\/hive_img_8791.jpg","image_full_path":"http:\/\/tlwarc.hg.gatech.edu\/\/sites\/default\/files\/images\/hive_img_8791.jpg","mime":"image\/jpeg","size":1006682,"path_740":"http:\/\/tlwarc.hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/images\/hive_img_8791.jpg?itok=9gNX0Ys7"}},"657143":{"id":"657143","type":"image","title":"The Summit supercomputing center at Oak Ridge National Laboratory. Credit: Oak Ridge National Laboratory.","body":null,"created":"1649684925","gmt_created":"2022-04-11 13:48:45","changed":"1649684925","gmt_changed":"2022-04-11 13:48:45","alt":"","file":{"fid":"249064","name":"2018-P01537.jpg","image_path":"\/sites\/default\/files\/images\/2018-P01537.jpg","image_full_path":"http:\/\/tlwarc.hg.gatech.edu\/\/sites\/default\/files\/images\/2018-P01537.jpg","mime":"image\/jpeg","size":49829,"path_740":"http:\/\/tlwarc.hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/images\/2018-P01537.jpg?itok=483xMUYJ"}}},"media_ids":["657354","657142","657144","657143"],"related_links":[{"url":"https:\/\/github.com\/FreshAirTonight\/af2complex","title":"Download AF2Complex"},{"url":"https:\/\/research.gatech.edu\/ai-tool-pairs-protein-pathways-clinical-side-effects-patient-comorbidities-suggest-targeted-covid","title":"AI Tool Pairs Protein Pathways with Clinical Side Effects, Patient Comorbidities to Suggest Targeted Covid-19 Treatments"}],"groups":[{"id":"1278","name":"College of Sciences"},{"id":"1275","name":"School of Biological Sciences"},{"id":"1188","name":"Research Horizons"}],"categories":[{"id":"135","name":"Research"},{"id":"138","name":"Biotechnology, Health, Bioengineering, Genetics"},{"id":"153","name":"Computer Science\/Information Technology and Security"},{"id":"146","name":"Life Sciences and Biology"}],"keywords":[{"id":"190340","name":"AlphaFold 2"},{"id":"175987","name":"protein structure"},{"id":"108061","name":"Oak Ridge National Laboratory"},{"id":"11937","name":"Jeffrey Skolnick"},{"id":"20381","name":"Mu Gao"},{"id":"9502","name":"Biomolecular"},{"id":"166882","name":"School of Biological Sciences"},{"id":"187915","name":"go-researchnews"},{"id":"187423","name":"go-bio"}],"core_research_areas":[{"id":"39441","name":"Bioengineering and Bioscience"},{"id":"39431","name":"Data Engineering and Science"}],"news_room_topics":[],"event_categories":[],"invited_audience":[],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[{"value":"\u003Cp\u003E\u003Cstrong\u003EWriter:\u0026nbsp;\u003C\/strong\u003E\u003Ca href=\u0022mailto:davidson.audra@gatech.edu\u0022\u003EAudra Davidson\u003C\/a\u003E\u003Cbr \/\u003E\r\nCommunications Officer\u003Cbr \/\u003E\r\nCollege of Sciences at Georgia Tech\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003EEditor:\u0026nbsp;\u003C\/strong\u003EJess Hunt-Ralston\u003Cbr \/\u003E\r\nDirector of Communications\u003Cbr \/\u003E\r\nCollege of Sciences at Georgia Tech\u003C\/p\u003E\r\n","format":"limited_html"}],"email":["davidson.audra@gatech.edu"],"slides":[],"orientation":[],"userdata":""}}}