In order to reduce the time associated with and the costs of drug discovery, machine learning is being used to automate much of the work in this process. However the size and complex nature of molecular data makes the application of machine learning especially challenging. Much work must go into the process of engineering features that are then used to train machine learning models, costing considerable amounts of time and requiring the knowledge of domain experts to be most effective. The purpose of this work is to demonstrate data driven approaches to perform the feature selection and extraction steps in order to decrease the amount of expert knowledge required to model interactions between proteins and drug molecules.

La possibilité d'utiliser une nouvelle méthodologie pour l'échange de liaisons peptidiques dans des conditions aérobiques et biocompatibles a été étudiée. Nous décrivons l'optimisation de la ligation chimique native combinatoire dynamique (dynNCL) au niveau d'un résidu N-méthylcystéine en utilisant des peptides modèles. Nous employons en outre cette méthode optimisée pour le criblage de bibliothèques de peptides combinatoires dynamiques en présence et en l'absence d'un gabarit de type anticorps. L'effet que l'incorporation d'une jonction dynamique au sein de ligands non-anticorps a sur l'affinité a également été étudié. Nous proposons que dynNCL peut être utilisé pour la conception d'une nouvelle classe de séquences pouvant potentiellement conduire au premier exemple d'épissage de protéines artificielles. / The possibility to use a new methodology for peptide bond exchange in aerobic biocompatible conditions has been investigated. We describe the assay optimization of dynamic combinatorial native chemical ligation (dynNCL) at the N-methyl-cysteine residue using model peptides. We further employ this optimized method for the screening of dynamic combinatorial peptide libraries in the presence and the absence of an antibody template. The effect that dynamic junction incorporation into non-antibody ligands has upon affinity was also studied. We propose that dynNCL can be used for the creation of a new class of designer sequences which can potentially provide the first example of artificial protein splicing.

The exponential growth of scientific literature is creating an increased need for systems to process and assimilate knowledge contained within text. Literature Based Discovery (LBD) is a well established field that seeks to synthesize new knowledge from existing literature, but it has remained primarily in the theoretical realm rather than in real-world application. This lack of real-world adoption is due in part to the difficulty of LBD, but also due to several solvable problems present in LBD today. Of these problems, the ones in most critical need of improvement are: (1) the over-generation of knowledge by LBD systems, (2) a lack of meaningful evaluation standards, and (3) the difficulty interpreting LBD output. We address each of these problems by: (1) developing indirect relatedness measures for ranking and filtering LBD hypotheses; (2) developing a representative evaluation dataset and applying meaningful evaluation methods to individual components of LBD; (3) developing an interactive visualization system that allows a user to explore LBD output in its entirety. In addressing these problems, we make several contributions, most importantly: (1) state of the art results for estimating direct semantic relatedness, (2) development of set association measures, (3) development of indirect association measures, (4) development of a standard LBD evaluation dataset, (5) division of LBD into discrete components with well defined evaluation methods, (6) development of automatic functional group discovery, and (7) integration of indirect relatedness measures and automatic functional group discovery into a comprehensive LBD visualization system. Our results inform future development of LBD systems, and contribute to creating more effective LBD systems.

The need for clear and organizationally effective communications is necessary to maintain sustainability as competition increases. Current research has not addressed problems associated with senior managers' clarity and intent and the misinterpretation by midlevel managers of that intent, causing division managers to misinterpret the company plans. Unresolved miscommunications may lead to destructive subculture development. This mixed methods design focused on how to minimize the confusion that manifests between senior and midlevel management within diverse and decentralized decision support structures. The secondary purpose was to advocate for the identification of divisional misalignment and provide information for a tool to help senior level managers identify possible misalignment. Leader-member exchange theory and decision theory guided the research design associated with the study of middle-level managers (N = 220) whose companies were members of the local Rotary clubs and selected individual businesses in South Carolina. Results were analyzed with correlations, ANOVA, and regression. Results indicated that the independent variables of clarity, information delivery tools, mental frame, and the form of message did not statistically affect the decision-making processes of middle level managers in similar-sized businesses. However, the qualitative results suggested that the senior manager's clarity is related directly to the distinction between FYI and FYA forms of communication. Positive social change may result from the findings. The results could be used to improve decision makers' ability to communicate their organizational strategy to other managers, thus promoting sustained businesses success and employment in a community.

由於近年社群網路各種應用網站興起，像是Facebook、Twitter，等，相關議 題也逐漸受到討論，例如越來越多利用社群網路傳播訊息或者病毒式行銷的相關 研究。當我們能夠找出一個社群網路當中，習慣的傳播模式或者是傳播路徑，並 且能從中定位各種角色的重要性，進一步在社群網路中找出這些角色後，在這些 相關的議題的應用將更加靈活。 目前各大社群網路應用網站，使用者都可以與社群網路中的好友分享自己的 動作，例如發佈影片或圖片，評論，按「讚」等，基於這樣的前提使用者的任何 活動是有機會被社群網路中的好友影響，因此我們定義好友間影響的可能性，以 及依觀察合理的定義出社群網路中較為重要的角色。 我們的演算法經由收集使用者在固定社群網路應用網站的各種動作，加上動 作的時間所形成的動作誌(action log)，以及使用者們所構成的社群網路，可以從 社群網路中找出主要的資訊傳遞路徑以及各種不同限制下的領袖以及追隨者，並 且將會利用社群網路應用網站驗證分析我們所定義的角色成為結論。 / Recently, social networking services and websites such as Facebook and Twitter are taking more and more parts in our daily life. Issues of in uence propagation have been studied in recent years. To ll in the gap of previous works, we aim to discover the main path of in uence and dene the importance of leader in hierarchy on the social graph. Social networking users are in uenced by the power of social networking service as they are able to post and likevideos, pictures and comments. Therefore, in this study we propose to discover the possibility of a relation and important roles by mining social activities. After collecting performed action and time stamp from dierent users and understanding their social network, our framework was able to identify the main in uence paths and leaders under dierent constrains. Most importantly, our approach outperforms both on precision/recall and ranking in realistic data.

This work takes place in the framework of Knowledge Discovery in Databases (KDD), often called "Data Mining". This domain is both a main research topic and an application ¯eld in companies. KDD aims at discovering previously unknown and useful knowledge in large databases. In the last decade many researches have been published about association rules, which are frequently used in data mining. Association rules, which are implicative tendencies in data, have the advantage to be an unsupervised model. But, in counter part, they often deliver a large number of rules. As a consequence, a postprocessing task is required by the user to help him understand the results. One way to reduce the number of rules - to validate or to select the most interesting ones - is to use interestingness measures adapted to both his/her goals and the dataset studied. Selecting the right interestingness measures is an open problem in KDD. A lot of measures have been proposed to extract the knowledge from large databases and many authors have introduced the interestingness properties for selecting a suitable measure for a given application. Some measures are adequate for some applications but the others are not. In our thesis, we propose to study the set of interestingness measure available in the literature, in order to evaluate their behavior according to the nature of data and the preferences of the user. The ¯nal objective is to guide the user's choice towards the measures best adapted to its needs and in ¯ne to select the most interesting rules. For this purpose, we propose a new approach implemented in a new tool, ARQAT (Association Rule Quality Analysis Tool), in order to facilitate the analysis of the behavior about 40 interest- ingness measures. In addition to elementary statistics, the tool allows a thorough analysis of the correlations between measures using correlation graphs based on the coe±cients suggested by Pear- son, Spearman and Kendall. These graphs are also used to identify the clusters of similar measures. Moreover, we proposed a series of comparative studies on the correlations between interestingness measures on several datasets. We discovered a set of correlations not very sensitive to the nature of the data used, and which we called stable correlations. Finally, 14 graphical and complementary views structured on 5 levels of analysis: ruleset anal- ysis, correlation and clustering analysis, most interesting rules analysis, sensitivity analysis, and comparative analysis are illustrated in order to show the interest of both the exploratory approach and the use of complementary views.

<p>This thesis covers design and implementation of a media system solution for home networks with personal computers and a set-top box. </p><p>In a home there are effectively two independent media systems with the same purpose: the personal computer and the digital set-top box, with the purpose of delivering digital media in form of audio and video to the consumer.</p><p>The goal of the thesis work was to implement a solution that bridges the gap between the two systems, so that the user, from the set-top box, can play back media that is actually stored on one or several personal computers. </p><p>Our solution is based on UPnP technology, which is used for service discovery and control. The choice of UPnP is motivated by an evaluation of discovery protocols, which concludes that UPnP is the most suitable technology in this particular system. Also, an evaluation of suitable transport protocols was done. Here,HTTP was used. </p><p>For the personal computer, a media server and a graphical user interface for configuring the media server were created. For the set-top box, a media client, and a graphical user interface for browsing the content of the media server, were created. In conclusion, the creation of the prototype was successful and the set-top box was able to playback media that had been shared by the PC on the network.</p>

<p>This thesis is focused on enzymatic studies of the secreted aspartic proteases (Saps) from <i>Candida albicans</i> as a tool for discovery of anti-<i>candida</i> drugs. <i>C. albicans</i> causes infections in a number of different locations, which differ widely in the protein substrates available and pH. Since <i>C. albicans</i> needs Saps during virulent growth, these enzymes are good targets for drug development.</p><p>In order to investigate the catalytic characteristics of Saps and their inhibitor affinities, substrate-based kinetic assays were developed. Due to the low sensitivity of these assays, especially at the sub-optimal pH required to mimic the different locations of infections, these assays were not satisfactory. Therefore, a biosensor assay was developed whereby, it was possible to study interaction between Saps and inhibitors without the need to optimise catalytic efficacy. Furthermore, the biosensor assay allowed determination of affinity, as well as the individual association and dissociation rates for inhibitor interactions.</p><p>Knowledge about substrate specificity, Sap subsite adaptivity, and the pH dependencies of catalytic efficacy has been accumulated. Also, screening of transition-state analogue inhibitors designed for HIV-1 protease has revealed inhibitors with affinity for Saps. Furthermore, the kinetics and pH dependencies of their interaction with Saps have been investigated. One of these inhibitors, BEA-440, displayed a complex interaction with Saps, indicating a conformational change upon binding and a very slow dissociation rate. A time dependent interaction was further supported by inhibition measurements. The structural information obtained affords possibilities for design of new more potent inhibitors that might ultimately become drugs against candidiasis. The strategy to combine substrate specificity studies with inhibitor screening has led to complementary results that generate a framework for further development of potent inhibitors.</p>

Landmark-based architecture has been commonly adopted in the networking community as a mechanism to measure and characterize a host's location on the Internet. In most existing landmark based approaches, end hosts use the distance measurements to a common, fixed set of landmarks to derive an estimated location on the Internet. This paper investigates whether it is possible for participating peer nodes in an overlay network to collaboratively construct an accurate geometric model of its topology in a completely decentralized peer-to-peer fashion, without using a fixed set of landmarks. We call such a peer-to-peer approach in topology discovery and modeling using landmarks PALM (Peers As LandMarks). We evaluate the performance characteristics of such a decentralized coordinates-based approach under several factors, including dimensionality of the geometric space, peer distance distribution, and the number of peer-to-peer distance measurements used. We evaluate two PALM-based schemes: RAND-PALM and ISLAND. In RAND-PALM, a peer node randomly selects from existing peer nodes as its landmarks. In ISLAND (Intelligent Selection of Landmarks), each peer node selects its landmarks by exploiting the topological information derived based on existing peer nodes' coordinates values. / Singapore-MIT Alliance (SMA)

The key to the complexity of higher organisms lies not in the number of protein coding genes they carry, but rather in the intrinsic complexity of the gene regulatory networks. The major effectors of transcriptional regulation are proteins called transcription factors, and in this thesis four papers describing genome-wide studies of seven such factors are presented, together with studies on components of the chromatin and transcriptome. In Paper I, we optimized a large-scale in vivo method, ChIP-chip, to study protein – DNA interactions using microarrays. The metabolic-disease related transcription factors USF1, HNF4a and FOXA2 were studied in 1 % of the genome, and a surprising number of binding sites were found, mostly far from annotated genes. In Paper II, a novel sequencing based method, ChIP-seq, was applied to FOXA2, HNF4a and GABPa, allowing a true genome-wide view of binding sites. A large overlap between the datasets were seen, and molecular interactions were verified in vivo. Using a ChIP-seq specific motif discovery method, we identified both the expected motifs and several for co-localized transcription factors. In Paper III, we identified and studied a novel transcription factor, ZBED6, using the ChIP-seq method. Here, we went from one known binding site to several hundred sites throughout the mouse genome. Finally, in Paper IV, we studied the chromatin landscape by deep sequencing of nucleosomal DNA, and further used RNA-sequencing to quantify expression levels, and extended the knowledge about the binding profiles for the transcription factors NFY and TCF7L2.