Uziela K.,University of Stockholm |
Wallner B.,Linkoping University |
Wallner B.,Swedish e Science Research Center
Bioinformatics | Year: 2016
Motivation: Model quality assessment programs are used to predict the quality of modeled protein structures. They can be divided into two groups depending on the information they are using: ensemble methods using consensus of many alternative models and methods only using a single model to do its prediction. The consensus methods excel in achieving high correlations between prediction and true quality measures. However, they frequently fail to pick out the best possible model, nor can they be used to generate and score new structures. Single-model methods on the other hand do not have these inherent shortcomings and can be used both to sample new structures and to improve existing consensus methods. Results: Here, we present an implementation of the ProQ2 program to estimate both local and global model accuracy as part of the Rosetta modeling suite. The current implementation does not only make it possible to run large batch runs locally, but it also opens up a whole new arena for conformational sampling using machine learned scoring functions and to incorporate model accuracy estimation in to various existing modeling schemes. ProQ2 participated in CASP11 and results from CASP11 are used to benchmark the current implementation. Based on results from CASP11 and CAMEO-QE, a continuous benchmark of quality estimation methods, it is clear that ProQ2 is the single-model method that performs best in both local and global model accuracy. © The Author 2016. Published by Oxford University Press.
Popovic J.,KTH Royal Institute of Technology |
Runborg O.,Swedish e Science Research Center
BIT Numerical Mathematics | Year: 2011
We propose and analyze a fast method for computing the solution of the high frequency Helmholtz equation in a bounded one-dimensional domain with a variable wave speed function. The method is based on wave splitting. The Helmholtz equation is split into one-way wave equations with source functions which are solved iteratively for a given tolerance. The source functions depend on the wave speed function and on the solutions of the one-way wave equations from the previous iteration. The solution of the Helmholtz equation is then approximated by the sum of the one-way solutions at every iteration. To improve the computational cost, the source functions are thresholded and in the domain where they are equal to zero, the one-way wave equations are solved with geometrical optics with a computational cost independent of the frequency. Elsewhere, the equations are fully resolved with a Runge-Kutta method. We have been able to show rigorously in one dimension that the algorithm is convergent and that for fixed accuracy, the computational cost is asymptotically just O(ω 1/p) for a pth order Runge-Kutta method, where ω is the frequency. Numerical experiments indicate that the growth rate of the computational cost is much slower than a direct method and can be close to the asymptotic rate. © 2011 Springer Science + Business Media B.V.
Koenigk T.,Swedish Meteorological and Hydrological Institute |
Koenigk T.,University of Stockholm |
Koenigk T.,Swedish e Science Research Center |
Caian M.,Swedish Meteorological and Hydrological Institute |
And 2 more authors.
Climate Dynamics | Year: 2016
Seasonal prediction skill of winter mid and high northern latitudes climate from sea ice variations in eight different Arctic regions is analyzed using detrended ERA-interim data and satellite sea ice data for the period 1980–2013. We find significant correlations between ice areas in both September and November and winter sea level pressure, air temperature and precipitation. The prediction skill is improved when using November sea ice conditions as predictor compared to September. This is particularly true for predicting winter NAO-like patterns and blocking situations in the Euro-Atlantic area. We find that sea ice variations in Barents Sea seem to be most important for the sign of the following winter NAO—negative after low ice—but amplitude and extension of the patterns are modulated by Greenland and Labrador Seas ice areas. November ice variability in the Greenland Sea provides the best prediction skill for central and western European temperature and ice variations in the Laptev/East Siberian Seas have the largest impact on the blocking number in the Euro-Atlantic region. Over North America, prediction skill is largest using September ice areas from the Pacific Arctic sector as predictor. Composite analyses of high and low regional autumn ice conditions reveal that the atmospheric response is not entirely linear suggesting changing predictive skill dependent on sign and amplitude of the anomaly. The results confirm the importance of realistic sea ice initial conditions for seasonal forecasts. However, correlations do seldom exceed 0.6 indicating that Arctic sea ice variations can only explain a part of winter climate variations in northern mid and high latitudes. © 2015, The Author(s).
Light S.,University of Stockholm |
Basile W.,University of Stockholm |
Elofsson A.,University of Stockholm |
Elofsson A.,Swedish e Science Research Center
Current Opinion in Structural Biology | Year: 2014
The frequency of de novo creation of proteins has been debated. Early it was assumed that de novo creation should be extremely rare and that the vast majority of all protein coding genes were created in early history of life. However, the early genomics era lead to the insight that protein coding genes do appear to be lineage-specific. Today, with thousands of completely sequenced genomes, this impression remains. It has even been proposed that the creation of novel genes, a continuous process where most de novo genes are short-lived, is as frequent as gene duplications. There exist reports with strongly indicative evidence for de novo gene emergence in many organisms ranging from Bacteria, sometimes generated through bacteriophages, to humans, where orphans appear to be overexpressed in brain and testis. In contrast, research on protein evolution indicates that many very distantly related proteins appear to share partial homology. Here, we discuss recent results on de novo gene emergence, as well as important technical challenges limiting our ability to get a definite answer to the extent of de novo protein creation. © 2014 Elsevier Ltd.
Wallner B.,Linkoping University |
Wallner B.,Swedish e Science Research Center
Bioinformatics | Year: 2014
Summary: Model Quality Assessment Programs (MQAPs) are used to predict the quality of modeled protein structures. These usually use two approaches: methods using consensus of many alternative models and methods requiring only a single model to do its prediction. The consensus methods are useful to improve overall accuracy; however, they frequently fail to pick out the best possible model and cannot be used to generate and score new structures. Single-model methods, on the other hand, do not have these inherent shortcomings and can be used to both sample new structures and improve existing consensus methods. Here, we present ProQM-resample, a membrane protein-specific single-model MQAP, that couples side-chain resampling with MQAP rescoring by ProQM to improve model selection. The side-chain resampling is able to improve side-chain packing for 96% of all models, and improve model selection by 24% as measured by the sum of the Z-score for the first-ranked model (from 25.0 to 31.1), even better than the state-of-the-art consensus method Pcons. The improved model selection can be attributed to the improved side-chain quality, which enables the MQAP to rescue good backbone models with poor side-chain packing. Availability and implementation: http://proqm.wallnerlab.org/download/. © The Author 2014.