Visual Clustering of Transcriptomic Data from Primary and Metastatic Tumors – Dependencies and Novel Pitfalls

1.AbstractPersonalized Oncology is a rapidly evolving area and offers cancer patients therapy options more specific than ever. Yet, there is still a lack of understanding regarding transcriptomic similarities or differences of metastases and corresponding primary sites. Approaching this question, we used two different unsupervised dimension reduction methods – t-SNE and UMAP – on three different metastases datasets – prostate cancer, neuroendocrine prostate cancer, and skin cutaneous melanoma – including 682 different samples, with three different underlying data transformations – unprocessed FPKM values, log10 transformed FPKM values, and log10+1 transformed FPKM values – to visualize potential underlying clusters. The approaches resulted in formation of different clusters that were independent of respective resection sites. Additionally, data transformation critically affected cluster formation in most cases. Of note, our study revealed no tight link between the metastasis resection site and specific transcriptomic features. Instead, our analysis demonstrates the dependency of cluster formation on the underlying data transformation and the dimension reduction method applied. These observations propose data transformation as another key element in the interpretation of visual clustering approaches apart from well-known determinants such as initialization and parameters. Furthermore, the results show the need for further evaluation of underlying data alterations based on the biological question and subsequently used methods and applications.

Visual Clustering Analysis of some traditional Mango (Mangifera indica L.) varieties of Murshidabad District, West Bengal using Clust Vis web tool

A clustergram or a heatmap is one of several techniques that directly visualize data without the need for dimensionality reduction. Heatmap is a representation of data in the form of a map or diagram in which data values are represented as colours. Cluster heatmaps have high data density, allowing them to compact large amounts of information into a small space. “ClustVis”, is a web tool for visualizing clustering of multivariate data using Principal Component Analysis and Heatmap. Using this web tool, genetic relationships among the traditional mango (Mangifera indica L.) varieties can be visualized. In this investigation ten (10) indigenous mango varieties were selected. These were elite varieties of Murshidabad viz. Anaras, Bhabani, Champa, Dilpasand, Kalabati, Kohinoor, Kohitoor, Molamjam. The morphological and biological characters were analyzed using this tool. Analysis and assessment of the current status of mango genetic resources will be important for ascertaining the relationship among traditional varieties. This data may be used for appropriate conservation and sustainable utilization measures. This information may also be needed to carry out breeding programs to develop improved cultivars for sustainable livelihoods of local communities.

Driver Liability Assessment in Vehicle Collisions in Spain

An accurate estimation of exposure is essential for road collision rate estimation, which is key when evaluating the impact of road safety measures. The quasi-induced exposure method was developed to estimate relative exposure for different driver groups based on its main hypothesis: the not-at-fault drivers involved in two-vehicle collisions are taken as a random sample of driver populations. Liability assignment is thus crucial in this method to identify not-at-fault drivers, but often no liability labels are given in collision records, so unsupervised analysis tools are required. To date, most researchers consider only driver and speed offences in liability assignment, but an open question is if more information could be added. To this end, in this paper, the visual clustering technique of self-organizing maps (SOM) has been applied to better understand the multivariate structure in the data, to find out the most important variables for driver liability, analyzing their influence, and to identify relevant liability patterns. The results show that alcohol/drug use could be influential on liability and further analysis is required for disability and sudden illness. More information has been used, given that a larger proportion of the data was considered. SOM thus appears as a promising tool for liability assessment.

Landscape Planning of Infrastructure through Focus Points’ Clustering Analysis. Case Study: Plastiras Artificial Lake (Greece)

Even though landscape quality is largely a subjective issue, the integration of infrastructure into landscapes has been identified as a key element of sustainability. In a spatial planning context, the landscape impacts that are generated by infrastructures are commonly quantified through visibility analysis. In this study, we develop a new method of visibility analysis and apply it in a case study of a reservoir (Plastiras dam in Greece). The methodology combines common visibility analysis with a stochastic tool for visual-impacts evaluation; points that generate high visual contrasts in landscapes are considered Focus Points (FPs) and their clustering in landscapes is analyzed trying to answer two questions: (1) How does the clustering of Focus Points (FPs) impact the aesthetic value of the landscape? (2) How can the visual impacts of these FPs be evaluated? Visual clustering is calculated utilizing a stochastic analysis of generated Zones of Theoretical Visibility. Based on the results, we argue that if the visual effect of groups of FPs is positive, then the optimal sitting of FPs should be in the direction of faint clustering, whereas if the effect is negative, the optimal sitting of FPs should be directed to intense clustering. In order to optimize the landscape integration of infrastructure, this method could be a useful analytical tool for environmental impact assessment or a monitoring tool for a project’s managing authorities. This is demonstrated through the case study of Plastiras’ reservoir, where the clustering of positively perceived FPs is found to be an overlooked attribute of its perception as a highly sustainable infrastructure project.

Smart cities in China: growth driver in the post pandemic world

Research background: In recent decades, a global trend towards the introduction of IoT technologies, artificial intelligence, blockchain, and many others into the field of urban management to create a single digital ecosystem has become increasingly noticeable. The urgency of this problem also manifested itself during the COVID-19 pandemic, and many components of the smart city made it possible to control and contain the spread of the infection. All these factors testify that in current conditions, the digitalization of cities is simply inevitable. In this regard, the number of smart cities continues to increase worldwide, and their development models are constantly improving under the influence of a considerable number of innovative solutions Purpose of the article: Of particular interest are the Chinese successes in the rapid digitalization of the economy and society and the increase in the number of smart cities. The study aims to analyze and identify trends in the development of smart cities in China. Methods: Since the concept of "smart city" is relatively modern and is in constant development, the authors have studied various articles and reports on this topic to identify different opinions about this topical problem. As part of the study, the authors carried out a visual clustering analysis of smart cities distribution in China. Findings & Value added: The authors were able to identify that the Chinese smart cities market has its own characteristic features, which allowed it to achieve such success. In more detail in this article, the authors focused on cluster development and highlighted the geographical pattern of smart cities distribution in China and their strengths and weaknesses in each area.

Mapping the Space of Chemical Reactions using Attention-Based Neural Networks

<div><div><div><p>Organic reactions are usually assigned to classes grouping reactions with similar reagents and mechanisms. Reaction classes facilitate communication of complex concepts and efficient navigation through chemical reaction space. However, the classification process is a tedious task, requiring the identification of the corresponding reaction class template via annotation of the number of molecules in the reactions, the reaction center and the distinction between reactants and reagents. In this work, we show that transformer-based models can infer reaction classes from non-annotated, simple text-based representations of chemical reactions. Our best model reaches a classification accuracy of 98.2%. We also show that the learned representations can be used as reaction fingerprints which capture fine-grained differences between reaction classes better than traditional reaction fingerprints. The unprecedented insights into chemical reaction space enabled by our learned fingerprints is illustrated by an interactive reaction atlas providing visual clustering and similarity searching. </p><p><br></p><p>Code: https://github.com/rxn4chemistry/rxnfp</p><p>Tutorials: https://rxn4chemistry.github.io/rxnfp/</p><p>Interactive reaction atlas: https://rxn4chemistry.github.io/rxnfp//tmaps/tmap_ft_10k.html</p></div></div></div>

Mapping the Space of Chemical Reactions using Attention-Based Neural Networks

<div><div><div><p>Organic reactions are usually assigned to classes grouping reactions with similar reagents and mechanisms. Reaction classes facilitate communication of complex concepts and efficient navigation through chemical reaction space. However, the classification process is a tedious task, requiring the identification of the corresponding reaction class template via annotation of the number of molecules in the reactions, the reaction center and the distinction between reactants and reagents. In this work, we show that transformer-based models can infer reaction classes from non-annotated, simple text-based representations of chemical reactions. Our best model reaches a classification accuracy of 98.2%. We also show that the learned representations can be used as reaction fingerprints which capture fine-grained differences between reaction classes better than traditional reaction fingerprints. The unprecedented insights into chemical reaction space enabled by our learned fingerprints is illustrated by an interactive reaction atlas providing visual clustering and similarity searching. </p><p><br></p><p>Code: https://github.com/rxn4chemistry/rxnfp</p><p>Tutorials: https://rxn4chemistry.github.io/rxnfp/</p><p>Interactive reaction atlas: https://rxn4chemistry.github.io/rxnfp//tmaps/tmap_ft_10k.html</p></div></div></div>