There exist many real-world applications such as recommendation systems, document retrieval, and computational biology where the correct ordering of instances is of equal or greater importance than predicting the exact value of some discrete or continuous outcome. Learning-to-Rank (LTR) refers to a group of algorithms that apply machine learning techniques to tackle these ranking problems. Despite their empirical success, most existing LTR models are not built to be robust to errors in labeling or annotation, distributional data shift, or adversarial data perturbations. To fill this gap, we develop four LTR frameworks that are robust to various types of perturbations. First, Pairwise Elastic Net Regression Ranking (PENRR) is an elastic-net-based regression method for drug sensitivity prediction. PENRR infers robust predictors of drug responses from patient genomic information. The special design of this model (comparing each drug with other drugs in the same cell line and comparing that drug with itself in other cell lines) significantly enhances the accuracy of the drug prediction model under limited data. This approach is also able to solve the problem of fitting on the insensitive drugs that is commonly encountered in regression-based models. Second, Regression-based Ranking by Pairwise Cluster Comparisons (RRPCC) is a ridge-regression-based method for ranking clusters of similar protein complex conformations generated by an underlying docking program (i.e., ClusPro). Rather than using regression to predict scores, which would equally penalize deviations for either low-quality and high-quality clusters, we seek to predict the difference of scores for any pair of clusters corresponding to the same complex. RRPCC combines these pairwise assessments to form a ranked list of clusters, from higher to lower quality. We apply RRPCC to clusters produced by the automated docking server ClusPro and, depending on the training/validation strategy, we show. improvement by 24%–100% in ranking acceptable or better quality clusters first, and by 15%–100% in ranking medium or better quality clusters first. Third, Distributionally Robust Multi-Output Regression Ranking (DRMRR) is a listwise LTR model that induces robustness into LTR problems using the Distributionally Robust Optimization framework. Contrasting to existing methods, the scoring function of DRMRR was designed as a multivariate mapping from a feature vector to a vector of deviation scores, which captures local context information and cross-document interactions. DRMRR employs ranking metrics (i.e., NDCG) in its output. Particularly, we used the notion of position deviation to define a vector of relevance score instead of a scalar one. We then adopted the DRO framework to minimize a worst-case expected multi-output loss function over a probabilistic ambiguity set that is defined by the Wasserstein metric. We also presented an equivalent convex reformulation of the DRO problem, which is shown to be tighter than the ones proposed by the previous studies. Fourth, Inversion Transformer-based Neural Ranking (ITNR) is a Transformer-based model to predict drug responses using RNAseq gene expression profiles, drug descriptors, and drug fingerprints. It utilizes a Context-Aware-Transformer architecture as its scoring function that ensures the modeling of inter-item dependencies. We also introduced a new loss function using the concept of Inversion and approximate permutation matrices. The accuracy and robustness of these LTR models are verified through three medical applications, namely cluster ranking in protein-protein docking, medical document retrieval, and drug response prediction.

Detection and analysis of protein-ligand binding sites is an important area of research in drug discovery. The FTMap web server is an established computational method for detection of binding hot spots, or regions on the protein surface that contribute disproportionately to the ligand binding free energy. This body of work primarily focuses on the utilization and advancement of FTMap for the study of protein-ligand interactions and their applications to drug discovery. First, the driving forces behind why some proteins require compounds beyond Lipinski’s rule-of-five (bRo5) guidelines are evaluated for 37 protein targets. Three types of proteins are identified on the basis of their binding hot spots, described by FTMap, and their ligand binding affinity profiles. We describe the multifaceted motivations for bRo5 drug discovery for each group of targets, including increased binding affinity, improved selectivity, decreased toxicity, and decreased off-target effects. Second, the conservation of surface binding properties in protein models is evaluated, with particular emphasis on their utility in drug discovery. Here, the probe-binding locations determined by FTMap are used to generate a binding fingerprint, and the Pearson correlation between the binding fingerprint of an experimental structure and a predicted model indicates the level of surface property conservation, without any knowledge of the protein function a priori. This analysis was performed on the protein models submitted to the Critical Assessment of Techniques for Protein Structure Prediction (CASP) rounds 12 and 14, and results were correlated with well-established structure quality metrics. Third, development of the publicly-available FTMove web server (https://ftmove.bu.edu) is described for detection of binding sites and their respective strengths across multiple different conformations of a protein. FTMove was tested on 22 proteins with known allosteric binding sites, and reliably identified both the orthosteric and allosteric binding sites as highly ranked binding sites. The results yield important insight into the dynamics and druggability of such binding sites. Finally, high throughput affinity purified, mass spectrometry data is evaluated for identification of protein-metabolite interactions (PMIs) in Escherichia coli. A detailed search for known PMIs in both the Protein Data Bank and KEGG database is described, and the resulting curated sets of 21 recovered and 37 potentially novel PMIs in E. Coli are presented. Finally, high confidence novel PMIs were evaluated with the template-based small molecule docking program, LigTBM. / 2023-01-26T00:00:00Z

Retrocyclin RC-101, a θ-defensin with lectin-like properties, potently inhibits infection by many HIV-1 subtypes by binding to the heptad repeat (HR)-2 region of gp41 and preventing six-helix bundle formation. In the present study, we used in silico computational exploration to identify residues of HR2 that interacted with RC-101 and then analyzed the HIV-1 Sequence Database at LANL for residue variations in the HR1 and HR2 segments that could plausibly impart in vivo resistance. Docking RC-101 to gp41 peptides in silico confirmed its strong preference for HR2 over HR1, and implicated residues crucial for its ability to bind HR2. We mutagenized these residues in pseudotyped HIV-1 JR.FL reporter viruses, and subjected them to single round replication assays in the presence of 1.25-10ug/ml RC-101. Except for one mutant that was partially resistant to RC-101, the other pseudotyped viruses with single-site cationic mutations in HR2 manifested absent or impaired infectivity or retained wild-type susceptibility to RC-101. Overall, these data suggest that most mutations capable of rendering HIV-1 resistant to RC-101 will also exert deleterious effects on the ability of HIV-1 to initiate infections - an interesting and novel property for a potential topical microbicide.

Protein-protein interactions build large networks which are essential in understanding complex diseases. Due to limitations of experimental methodology there are problems with large amounts of false negative and positive interactions; and a large gap in the amount of known interactions and structurally determined interactions. By using computational methods these problems can be alleviated. In this thesis the quality of a newly developed pipeline (InterPred) were investigated for its ability to generate coarse interaction models and score them. This ability was investigated by performing docking experiments in Rosetta on models generated in InterPred. The results suggest that InterPred is highly successful in generating good starting points for docking proteins in silico and to distinguish the quality of models.

No description available.

The presented work in this report is a master’s thesis at the machine design track at KTH Royal Institute of Technology. The work was commissioned by Ovzon through Svea Teknik AB. The T6 antenna in Ovzons antenna terminal portfolio is as of now fully manual in terms of positioning against satellites. The problem with this is that if the operator is situated in a vehicle, they are required to exit the vehicle and place the antenna on solid ground, aim it, and then return to the vehicle. To cope with the need of manually placing and aiming the antenna, a solution which enables the operator to fully control the antenna from within a vehicle is therefore desired. This work aims to develop a conceptual design that lets the antenna be mounted and automatically aimed against satellites while following tight dimension requirements. The work centered around a concept generating- and evaluation process where different main functions were identified, and to which concepts were formulated by brainstorming. The concepts were mapped with the help of a Functions/Means-Tree and then later evaluated with Pugh’s matrices in order to find a suitable one. The result was two conceptual designs out of which one was chosen by Ovzon for manufacturing. The chosen concept consists of a docking station which is positioned in elevation using a stepper motor connected to a planetary precision gearbox and in azimuth using a direct drive rotary table. The antenna is placed in the docking station and secured into place using latches. The manufactured proof of concept was constructed out of primarily 3D-printed parts but also some that were turned in steel. The resulting concept proved to solve the challenges and given requirements while still being small in size and robust to use. / Följande arbete är ett examensarbete inom spåret maskinkonstruktion vid KTH, Kungliga Tekniska Högskolan i Stockholm. Arbetet utfördes på uppdrag av Ovzon genom Svea Teknik AB. En av produkterna som ingår i Ovzons produktsortiment av antennterminaler är T6-antennen. Den är nuvarande helt manuell när det kommer till att positionera och rikta in sig mot satelliter. Problemet med detta är att om antennoperatören sitter i ett fordon så måste denne gå ut från fordonet och placera antennen på solid mark för att sedan manuellt rikta in antennen mot satelliten och därefter återvända till fordonet. En lösning som låter operatören helt och hållet kontrollera antennen från fordonet istället för att manuellt placera och rikta in den är därför efterfrågat. Detta arbete har som avsikt att konstruera en konceptuell design som låter antennen bli monterad och automatiskt inpekad mot satelliter, samtidigt som den uppfyller strikta krav gällande storlek. Arbetet är centrerat kring en konceptgenererings- och utvärderingsprocess där olika huvudfunktioner hos enheten var identifierade och till vilka koncept var formulerade genom diskussion. Koncepten var sedan kartlagda med hjälp av en Function/Means-Tree och sedan utvärderade med hjälp av Pugh’s-matriser för att på så vis hitta den optimala lösningen. Resultatet var två konceptuella konstruktioner varav en blev vald av Ovzon för att bli tillverkad som prototyp. Det valda konceptet bestod av en dockningsstation som riktas i elevation med hjälp av en stegmotor som är kopplad till en precisionsplanetväxellåda och i azimut med hjälp ett direktdrivet rotationsbord. Antennen placeras i dockningsstationen och hålls fast med hjälp av spännen. Den tillverkade prototypen bestod till största del av additivt tillverkade komponenter men även några av svarvat stål. Det slutgiltiga konceptet visade sig lösa många av de initial autmaningarna och givna kraven medan den samtidigt var liten i storlek och robust att använda.

G protein-coupled receptors (GPCRs) represent the largest membrane protein family and a significant target class for therapeutics. Receptors from GPCRs’ largest class, class A, influence virtually every aspect of human physiology. About 45% of the members of this family endogenously bind flexible peptides or peptides segments within larger protein ligands. While many of these peptides have been structurally characterized in their solution state, the few studies of peptides in their receptor-bound state suggest that these peptides interact with a shared set of residues and undergo significant conformational changes. For the purpose of understanding binding dynamics and the development of peptidomimetic drug compounds, further studies should investigate the peptide ligands that are complexed to their cognate receptor.

The pathogenicity and intractable nature of the microorganism Staphylococcus aureus (SA) has been long documented and highlighted by many health care agencies, with emphasis on its ability to exploit the human coagulation system to deadly effect. Two drugs from a class of inhibitors known as Direct Thrombin Inhibitors (DTI) have been shown to have a substantial effect on the enzyme secreted by SA known as Staphylocoagulase (SC), but up until now the application of this potential treatment has been limited. This paper strives to supply an overview of these clinical studies and propose a novel protocol for testing DTI's on SA in an in vitro setting. Three DTIs have been identified, including two already tested in clinical trials, and computational molecular docking simulations have been applied to elucidate the mechanisms of action for the inhibition. An additional DTI has been developed using these mechanisms as principles and shows promise for future development. After conducting this preliminary protocol, it has been found that running a minimum inhibitory concentration test across several tubes with varying degrees of these DTIs demonstrated varying levels of coagulation consistent with the findings of clinical research papers. It is fair to conclude, then, that after development or discovery of new coagulase inhibitors, they can be quickly and accurately tested against existent DTIs to gauge their efficacy.

No description available.

No description available.