1 - 11 of 11 results

AUTO–MUTE / AUTOmated server for predicting functional consequences of amino acid MUTations in protEins

A collection of programs for predicting functional changes to proteins upon single residue substitutions, developed by combining structure-based features with trained statistical learning models. For each type of function prediction, a variety of classification and regression models have been developed and are available for researchers. These include Random Forest, Support Vector Machine (SVM), AdaBoostM1 combined with the C4.5 Decision Tree algorithm, as well as Tree and SVM regression. The trained classifiers provide instantaneous and reliable predictions regarding HIV-1 co-receptor usage, requiring only translated V3 loop genotypes as input. Furthermore, the novelty of these computational mutagenesis based predictor attributes distinguishes the models as orthogonal and complementary to previous methods that utilize sequence, structure, and/or evolutionary information.

CM Classifier

A web server for HIV coreceptor tropism determination. CM Classifier is composed of two coreceptor-specific weight matrices (CMs) based on a full-scale dataset. Our classifier outperformed all other comparable methods for independent dataset. Several studies indicated that general genotypic methods might lack accuracy for non-B subtypes. In order to address this issue, another two subtype specific classifiers were also developed for subtype C and D, which are also spreading rapidly and represent a large proportion of worldwide infections.


A computional coreceptor usage prediction model. gCUP is based on our recently developed method T-CUP, but was redeveloped, parallelized and optimized for the use on graphics processing units (GPUs). gCUP and T-CUP give identical predictions and thus the accuracy of the model is not compromised by using GPUs. By harvesting the power of GPUs and optimizing the use of their fast local memory, gCUP can drastically reduce the runtime and process the same 40 million reads in just 4 min using one modern GPU.


A tool for predicting HIV-1 coreceptor usage from the V3 region of the HIV envelope protein gp120. Geno2pheno analyzes the NGS data in a elaborate fashion and performs better than existing methods without training any parameters on the test data. Additionally, we show how one can obtain interpretable prediction results and evaluate information on which of the residues of the V3 loop contribute to the improvement of prediction accuracy. Specifically, we find amino acids at certain positions that are highly predictive and might lead to new insights about the interaction between the V3 loop and the different coreceptors.


A model for predicting co-receptor tropism. T-CUP 2.0 is based on our recently published T-CUP model. T-CUP 2.0 models co-receptor tropism using information of the electrostatic potential and hydrophobicity of V3-loops. We found that it is possible to model co-receptor tropism in HIV-1 based on a simplified structure-based model of the V3 loop. In this way, genotypic prediction of co-receptor tropism is very accurate, fast and can be applied to large datasets derived from next-generation sequencing technologies.


Permits management and downstream analysis of pathogen sequence data. Segminator quantifies the effects of divergence on the mapping of reads to a template sequence. It employs joint nucleotide frequencies across temporally sampled data sets from a single host to derive a set of posterior probabilities for each of the possible nucleotides at each site. This tool was used to characterize the emergence of low frequency CXCR4-using variants following treatment with an HIV entry-inhibitor drug.


A suite of highly accurate in silico tests for all of the major HIV subtypes, namely subtypes A, B, C, D, CRF01_AE and CRF02_AG, which together account for 95% of HIV-1 infections worldwide. PhenoSeq will inform the appropriate use of maraviroc in regions of the world where non-B HIV predominates, which are burdened the most by the HIV-1 pandemic. The PhenoSeq platform comprises a suite of subtype specific V3 sequence-based genotypic tropism tests, designed specifically to predict the coreceptor usage of HIV-1 subtype A, B, C, D, CRF01_AE and CRF02_AG viruses. PhenoSeq is an updated version of CoRSeqV3-C.

HIV-1 coreceptor usage prediction without sequence alignments

A string kernel (the distant segments kernel), a SVM predictor for HIV-1 coreceptor usage with the identification of the most relevant features and state-of-the-art results on accuracy, specificity, sensitivity and receiver operating characteristic. As suggested, string kernels outperform all published algorithms for HIV-1 coreceptor usage prediction. Large margin classifiers and string kernels promise improvements in drug selection, namely CCR5 coreceptor inhibitors and CXCR4 coreceptor inhibitors, in clinical settings. Since the binding of an envelope protein to a receptor/coreceptor prior to infection is not specific to HIV-1, one could extend this work to other diseases.