Posts by Dr. Armen
|
1)
Message boards : Docking@Home Science : What is the main goal of Docking@home ? ( Message 1198 )Posted 3936 days ago by 
Dr. Armen
I have a question. Suguruhirahara, Michela provided a great answer to this question. However, I will also add that the protein-ligand complex 1tng is part of one of our test-sets. In the course of our project, we will apply different docking models to try to dock the individual test-case 1tng many times to see how our models perform. Each time we make a change to our model, we will likely see how we perform on 1tng. I myself do not know how many protein-ligand complexes will be tested in the alpha, beta and full-operational phase of our project; the short answer is literally as many as possible. I have been working for months on constructing different test-sets that we will use to test various aspects of our docking methodology. I will keep you updated on our progress in the future. Right now, during the alpha phase, we are performing some preliminary studies and also doing things like error-checking to make sure that we are ready for prime-time. |
|
2)
Message boards : Docking@Home Science : What is the main goal of Docking@home ? ( Message 1082 )Posted 3941 days ago by
Dr. Armen
Over @Rosetta a question came up regarding the comparability of Docking and Rosetta. This thread, and especially Dr. Armens posts were igven as an explanation. As a general policy, I would prefer to not summarize what is done on other life-science BOINC projects, and I would direct you to the Find-a-Drug site ( http://www.find-a-drug.org ) or also to the wikipedia entry ( http://en.wikipedia.org/wiki/Find-a-drug ). To my knowledge, the Find-a-Drug site is now permanently closed as of Dec 2005. However, since we are a new site, I will provide a brief comparison of what has been done in their project and how this compares and contrasts with the scientific goals and methods of our project. Keith Davies ran the project Find-a-Drug and previously the CAN-DDO cancer screening project that was funded by the National Foundation for Cancer Research (NFCR). These projects used the proprietary software THINK (developed by Treweren Consultants - Keith Davies and coworkers) for virtual screening of protein drug targets for various diseases including cancer. Some of the results for the cancer drug targets were then validated using NCI cancer screening data. As there are many different methodological approaches to protein folding, there are also many different approaches to protein-ligand docking. The proprietary software THINK uses a "pharmacophore pattern matching" approach with a full conformational search. This method is able to search rapidly through very large databases of available compounds and as well as computer-generated derivates of available compounds (theoretical compounds). To my knowledge the THINK methodology has not been published in peer-reviewed scientific journal, but the method is summarized in the following publications: Davies, E. K., Glick, M., Harrison, K. N. & Richards, W. G. (2002). Pattern recognition and massively distributed computing. Journal of Computational Chemistry. 23(16): 1544-1550. Davies, E. K. & Richards, W. G. (2002). The potential of internet computing for drug discovery. Drug Discovery Today, 7(11 Suppl): S99-S103. Also, some details about the methods and results from the Find-a-Drug project are available in a pdf from the CCP4 website (Collaborative Computational Project Number 4 in Protein Crystallography). ( http://www.ccp4.ac.uk/maxinf/mm4mx/PDFs/K.Davies.pdf ) The methods that we use for protein-ligand docking are quite different that the pharmacophore pattern matching approach employed by THINK. Our method is likely to be much more accurate, but is also significantly more computationally expensive, and is therefore not as well-suited for searching for very large databases of compounds. We currently use a molecular mechanics all-atom CHARMM-based docking method with fully flexible ligands, and a rigid protein conformation. One of the goals of our project is to develop models to incorporate flexibility into the protein as well, although this is very challenging. The details of the methodology used in our CHARMM-based docking (also known as CDOCKER) and several benchmark tests of docking and scoring have been published in several peer reviewed journal articles (shown below). In terms of the general approach and scientific goals of the project, there are some notable differences between Find-a-Drug and Docking@home. The Find-a-Drug project aimed at virtual screening of very large databases of compounds for specific drug targets with the goal of identifying as many "hits" or drug-like molecules as possible that might lead to new drugs. This approach assumes that the method has sufficient accuracy for a large scale application of the method. We applaud the pioneering work of the Find-a-Drug project, and we also intend to do such large virtual screens in the future. However, we still have a lot of work to do before we can perform virtual screening on such a large scale. Our approach is that we need to spend our efforts on improving our method and its accuracy, and then apply it to specific applications. We also need to develop and improve our databases of compounds. Therefore, our primary scientific goals in the short term are to improve and validate our methodology so we can improve accuracy. This means that we have to carefully examine how our methods perform when we know the answers (benchmark tests). In this way, it is our goal to develop and validate new methods of incorporating protein flexibility into our docking method without compromising the accuracy of our predictions. For example, in a virtual screen it is known that not including protein flexibility reduces accuracy. This is because compounds that are known to bind may not be able to bind given a specific rigid conformation of the protein that does not perfectly accommodate them (these are known as false negatives). Alternatively, introducing protein flexibility in a virtual screen may allow compounds to score well that are known not to bind, or bind with a low affinity (these are known as false positives). If the false positive rate becomes to large, then introducing flexibility into a virtual screen will have little or no benefit. This is a paradox of introducing protein flexibility into protein ligand docking that we intend to study. It is our goal to study various models of incorporating protein flexibility into docking, and to assess their accuracy with regards to this paradox. For this reason, our bench-mark studies will focus on a set of proteins that are known to be flexible. We will also perform example virtual screens where the "answer is known" from experimental high-through-put screening, so we can directly assess the effect of protein flexibility on the number of false positives and false negatives. Finally, because we intend to publish all of our methods and results in peer-reviewed scientific journals, any advancements in methodology will be available to the general public and academic researchers. Selected peer-reviewed scientific journal articles: Taufer, M., Crowley, M., Price, D., Chien, A. A., & Brooks, C. L. III (2005). Study of an accurate and fast protein-ligand docking algorithm based on molecular dynamics. Concurrency and Computation: Practice and Experience. 17(14): 1627-1641. Erickson, J. A., Jalaie, M., Robertson., D. H., Lewis, R. A., & Vieth, M. (2004). Lessons in molecular recognition: the effects of ligand and protein flexibility on molecular docking accuracy. Journal of Medicinal Chemistry, 47(1): 45-55. Ferrara, P., Gohlke, H., Price, D. J., Klebe, G. & Brooks, C. L. III (2004). Assessing scoring functions for protein-ligand interactions. Journal of Medicinal Chemistry, 47(12): 3032-3047. Bursulaya, B. D., Totrov, M., Abagyan, R. & Brooks, C. L. III (2003). Comparative study of several algorithms for flexible ligand docking. Journal of Computer Aided Molecular Desing. 17(11): 755-763. Wu, G., Robertson, D. H., Brooks, C. L. III. & Vieth, M. (2003). Detailed analysis of grid-based molecular docking: A case study of CDOCKER- a CHARMm-based MD docking algorithm. Journal of Computational Chemistry, 24(13):1549-1562. Vieth, M., Hirst, J. D., Kolinski, A. & Brooks, C. L. III (1998). Assessing energy functions for flexible docking. Journal of Computational Chemistry, 19(14):1612-1622. Vieth, M., Hirst, J. D., Dominy, B. N., Daigler, H., & Brooks, C. L. III (1998). Assessing search strategies for flexible docking. Journal of Computational Chemistry, 19(14): 1623-1631. |
|
3)
Message boards : Docking@Home Science : What is the main goal of Docking@home ? ( Message 358 )Posted 3978 days ago by
Dr. Armen
Thanks for the answers, it is more obvious what Docking project is doing in term of science. Docking@Home is a pure non-profit effort, and we will not make money from the results. We are academic researchers that are funded by public agencies [NSF and NIH] so that we can benefit the public. This is accomplished by publishing our results and methods in peer reviewed scientific journals, releasing our results to the public via this project web site, and eventually by collaborating with other publicly funded experimental researchers. By collaborating with other publicly funded experimental researchers, we hope this will allow us all to get more out of our limited publicly funded research dollars. The only connection between Predictor and Docking is that our project leader Michela Taufer helped to establish the Predictor site in the Charlie Brooks III lab at TSRI, and is now establishing the Docking site in her own lab at UTEP. Predictor is a site for the prediction of protein structure, folding, and structural transitions. Docking is a site for the prediction of protein-ligand interactions using CHARMM based molecular docking methods. |
|
4)
Message boards : Docking@Home Science : What is the main goal of Docking@home ? ( Message 357 )Posted 3978 days ago by
Dr. Armen
Would the scientific results be free for the public ? Yes, scientific results will be freely availible for the public. Initially, we will most likely be the main user of our results, because we wish to test various models of protein flexibility for protein-ligand docking. So we will try a model and then see if it works to correctly predict known experimental results in difficult test cases where the protein is known to be flexible. Some models will perform better than others, and we will decide which results are likely to be "the most usefull" to other researchers in this area and then release them. Others researchers in this area can benifit from the release of our results. HIV protease is an excellent example of one of the most flexible proteins that will indeed be part of our study. |
|
5)
Message boards : Docking@Home Science : What is the main goal of Docking@home ? ( Message 250 )Posted 3980 days ago by
Dr. Armen
>>> I would like to thank you Dr Armen for your detailed response to my query. Your goals seem very worthwhile and I will enjoy crunching some data for you and your team and helping to sort out the problems of getting you the information that you require. Thanks Conan, I will do my best to continue to provide detailed responses to questions as soon as possible. Thanks for participating in our project we really apreciate it. |
|
6)
Message boards : Docking@Home Science : What is the main goal of Docking@home ? ( Message 248 )Posted 3980 days ago by
Dr. Armen
Thanks for explanation, very appreaciated. With regards to our immediate scientific goal, our results will be used to improve our protein-ligand docking methodology and to validate our models. Developing and validating our methods will improve the effectiveness of our application of protein-ligand docking to real-world scientific problems and future collaborations with experimentalists. As a policy, I would prefer to not summarize what is done on other life-science BOINC projects, and I would direct you all to look at the pages explaining the science of each individual project for specific details. However, because we are a new site, I will provide a brief comparison of the general types of work that is done on each site. Rosetta@Home, Folding@Home, Tanpaku, and Predictor@Home all perform predictions about protein structure and protein folding. Rosetta@Home also performs predictions for protein design and protein-protein interactions. Dr. David Baker does an excellent job of explaining the potential applications of protein folding, protein design and protein-protein interactions to fighting disease here: http://boinc.bakerlab.org/rosetta/rah_medical_relevance.php SIMAP@Home uses sequence information to determine protein similarities and protein domains. The prediction of domains is also used for the functional annotation of proteins. FightAIDS@Home performs protein-ligand docking predictions using AutoDock. Their site specifically targets important HIV virus proteins. They are currently investigating the structural basis for the evolution of drug resistance in important HIV drug targets, with the goal of identifying new drug-like small molecules that are less susceptible to drug resistance. Our site Docking@Home will also perform protein-ligand docking predictions, but we are using a CHARMM based approach to docking, not AutoDock. Our immediate scientific goals are to improve and validate our docking methodology, and then to apply it to specific scientific problems. In the initial stages of our site, we will work on a large "test set" of protein-ligand complexes that includes important protein drug targets for HIV/AIDS, cancer and other diseases. During the initial stages of our project we intend to use these examples because they are of high interest to the biomedical community, well-studied, and have lots of excellent experimental data that we can compare to and validate our methods. I urge all of you to participate in as many of the life-science BOINC projects as possible as they each provide an important and unique contribution to the rapidly progressing field of computational biology. Thank You, Dr. Roger S. Armen PhD |
|
7)
Message boards : Docking@Home Science : What is the main goal of Docking@home ? ( Message 209 )Posted 3981 days ago by
Dr. Armen
Does Docking@home mainly search for how proteins can lock onto (dock) with other proteins, viruses and bacteria to see how a particular protein can neutralise or aid some other protein/molecule to help fight diseases? Hello, my name is Dr. Roger S. Armen PhD, and I am the primary scientist directing the development and application of molecular docking techniques for the Docking@Home project. I am a NIH funded postdoctoral fellow in the Charlie Brooks III lab at The Scripps Research Institute in San Diego. My research focuses on the docking of small molecules to proteins using computational chemistry. Although it is true that proteins indeed dock to other proteins, this is not the primary focus of our project for the first few years. Our project will focus on the docking of small molecules (drug- like molecules also called "ligands") to proteins and developing and extending our current methodology for protein-ligand docking. Protein-ligand docking is now a very useful step in the identification of new small molecules that may bind to a protein. A protein-ligand docking simulation can help direct experimental investigations of important proteins of interest, such as proteins implicated in diseases. In principle, protein-ligand docking may aid in the identification of new "drug-like" small molecules that may be re-designed into molecules with more favorable "drug-like" properties. Therefore, protein-ligand docking is a general computational method that may be useful in the "structure-based-design" of new drug-like molecules, or as a tool for the design of protein-ligand interactions in general. The immediate scientific goals of our docking@home project are aimed at the development of our protein-ligand docking methodology. Our CHARMM based docking method has already been established as one of the most accurate docking methods that currently exist for docking of a flexible ligand to a rigid protein. We intend to continue to develop, extend, and validate our methods on new and more difficult test cases. We intend to develop accurate methods to include protein flexibility in protein-ligand docking, which is currently one of the most pressing issues in the field of docking. The validation of such an approach on a wide variety of protein targets would be a very important contribution to the scientific community. I received a postdoctoral fellowship from the NIH for my proposal to solve this important problem in computational biology. Our secondary scientific goal is to apply our protein-ligand docking methods to important scientific problems. Eventually, we would like to open our site to collaborations with publicly funded experimental scientists who have a specific need for accurate protein-ligand docking, but do not have the time, resources, expertise or personnel for such a project. Thank you very much for your interest in the project. Dr. Roger S. Armen PhD |