Marco Necci1,2, Damiano Piovesan1, Zsuzsanna Dosztányi3,4, Silvio C E Tosatto1,5. 1. Department of Biomedical Sciences, University of Padua, 35121 Padova, Italy. 2. Fondazione Edmund Mach, 38010 San Michele all'Adige, Italy. 3. MTA-ELTE Lendület Bioinformatics Research Group, Department of Biochemistry, Eötvös Loránd University, Budapest, Hungary. 4. Institute of Enzymology, Research Centre for Natural Sciences, Hungarian Academy of Sciences, H-1518 Budapest, Hungary. 5. CNR Institute of Neuroscience, 35121 Padova, Italy.
Abstract
Motivation: Intrinsic disorder (ID) is established as an important feature of protein sequences. Its use in proteome annotation is however hampered by the availability of many methods with similar performance at the single residue level, which have mostly not been optimized to predict long ID regions of size comparable to domains. Results: Here, we have focused on providing a single consensus-based prediction, MobiDB-lite, optimized for highly specific (i.e. few false positive) predictions of long disorder. The method uses eight different predictors to derive a consensus which is then filtered for spurious short predictions. Consensus prediction is shown to outperform the single methods when annotating long ID regions. MobiDB-lite can be useful in large-scale annotation scenarios and has indeed already been integrated in the MobiDB, DisProt and InterPro databases. Availability and Implementation: MobiDB-lite is available as part of the MobiDB database from URL: http://mobidb.bio.unipd.it/. An executable can be downloaded from URL: http://protein.bio.unipd.it/mobidblite/. Contact: silvio.tosatto@unipd.it. Supplementary information: Supplementary data are available at Bioinformatics online.
Motivation: Intrinsic disorder (ID) is established as an important feature of protein sequences. Its use in proteome annotation is however hampered by the availability of many methods with similar performance at the single residue level, which have mostly not been optimized to predict long ID regions of size comparable to domains. Results: Here, we have focused on providing a single consensus-based prediction, MobiDB-lite, optimized for highly specific (i.e. few false positive) predictions of long disorder. The method uses eight different predictors to derive a consensus which is then filtered for spurious short predictions. Consensus prediction is shown to outperform the single methods when annotating long ID regions. MobiDB-lite can be useful in large-scale annotation scenarios and has indeed already been integrated in the MobiDB, DisProt and InterPro databases. Availability and Implementation: MobiDB-lite is available as part of the MobiDB database from URL: http://mobidb.bio.unipd.it/. An executable can be downloaded from URL: http://protein.bio.unipd.it/mobidblite/. Contact: silvio.tosatto@unipd.it. Supplementary information: Supplementary data are available at Bioinformatics online.
Authors: Jie Wang; Jeong-Mo Choi; Alex S Holehouse; Hyun O Lee; Xiaojie Zhang; Marcus Jahnel; Shovamayee Maharana; Régis Lemaitre; Andrei Pozniakovsky; David Drechsel; Ina Poser; Rohit V Pappu; Simon Alberti; Anthony A Hyman Journal: Cell Date: 2018-06-28 Impact factor: 41.582