Quantitative Structure-Activity Relationship (QSAR) models are critical in various areas of drug discovery, for example in lead optimisation and virtual screening. Recently, the need for models that are not only predictive but also interpretable has been highlighted. In this paper, a new methodology is proposed to build interpretable QSAR models by combining elements of network analysis and piecewise linear regression. The algorithm presented, modSAR, splits data using a two-step procedure. First, compounds associated with a common target are represented as a network in terms of their structural similarity, revealing modules of similar chemical properties. Second, each module is subdivided into subsets (regions), each of which is modelled by an independent linear equation. Comparative analysis of QSAR models across five data sets of protein inhibitors obtained from ChEMBL is reported and it is shown that modSAR offers similar predictive accuracy to popular algorithms, such as Random Forest and Support Vector Machine. Moreover, we show that models built by modSAR are interpretatable, capable of evaluating the applicability domain of the compounds and serve well tasks such as virtual screening and the development of new drug leads.
Networks constitute powerful means of representing various types of complex systems, where nodes denote the system entities and edges express the interactions between the entities. An important topological property in complex networks is community structure, where the density of edges within subgraphs is much higher than across different subgraphs. Each of these subgraphs forms a community (or module). In literature, a metric called modularity is defined that measures the quality of a partition of nodes into different mutually exclusive communities. One means of deriving community structure is modularity maximisation. In this paper, a novel mathematical programming-based model, DiMod, is proposed that tackles the problem of maximising modularity for directed networks.
The LSE Public Policy Review is an open-access, refereed journal which is published quarterly. Each issue is thematic and concentrates on a key topic at the heart of current debates in public policy. Public policy challenges bring to the fore cross-cutting questions which require a global perspective and a focus on their interconnectedness. Because of this, articles in each issue take different disciplinary perspectives, encouraging interdisciplinary collaboration and analysis at the forefront of current thinking. As a result, each issue presents a comprehensive approach to the specific theme and an analysis that is academically rigorous but also readily accessible to all readers. The LSEPPR publishes original research papers, conceptual articles, review papers written for a general readership, in non-technical language aimed at a wide audience including government, business and policy-makers, as well as academics and students. LSEPPR seeks to actively contribute to the study and development of public and social policy, public administration and public management.
15 February 2016 in The European Physical Journal B
Atualmente, o grande volume de notícias falsas circulando nas redes sociais é um perigo para a percepção da realidade da sociedade. Técnicas de aprendizado de máquina têm sido úteis para o combate da desinformação, mas para gerar bons resultados necessitam de conjuntos de dados de treinamento balanceados e com boa qualidade. Uma vez que os principais corpus para o treinamento de modelos detecção de notícias falsas disponíveis publicamente estão desatualizados ou desalinhados, neste trabalho foi proposto um enfoque inovador para recuperar as notícias verdadeiras a partir das falsas, e melhorar sua similaridade e alinhamento. Desta forma, foi desenvolvido um conjunto de dados que permite verificar e classificar a informação que consumimos diariamente na rede, por meio de processamento da linguagem natural. Além disso, o corpus resultante foi avaliado utilizando técnicas clássicas de processamento da linguagem natural para a representação de textos, como BoW e BoW TF-IDF, juntamente com vários métodos tradicionais de classificação. Os resultados obtidos demonstram que este conjunto de dados é efetivo para a classificação de notícias, com um f1-score de 0.945 usando um classificador de Multi-layer perceptron. Portanto, este novo corpus configura-se como um recurso valioso na luta contra a desinformação e para melhorar a qualidade da informação disponível online.