Integrating convolution and self-attention improves language model of human genome for interpreting non-coding regions at base-resolution

Interpretation of non-coding genome remains an unsolved challenge in human genetics due to impracticality of exhaustively annotate biochemically active elements in all conditions. Deep learning based computational approaches emerge recently to help interpretating non-coding regions. Here we present LOGO (Language of Genome), a self-attention based contextualized pre-trained language model containing only 2 self-attention layers with 1 million parameters as a substantially light architecture that applies self-supervision techniques to learn bidirectional representations of unlabeled human reference genome. LOGO is then fine-tuned for sequence labelling task, and further extended to variant prioritization task via a special input encoding scheme of alternative alleles followed by adding a convolutional module. Experiments show that LOGO achieves 15% absolute improvement for promoter identification and up to 4.5% absolute improvement for enhancer-promoter interaction prediction. LOGO exhibits state-of-the-art multi-task predictive power on thousands of chromatin features with only 3% parameterization benchmarking against fully supervised model, DeepSEA and 1% parameterization against a recent BERT-based language model for human genome. For allelic-effect prediction, locality introduced by one dimensional convolution shows improved sensitivity and specificity for prioritizing non-coding variants associated with human diseases. In addition, we apply LOGO to interpret type 2 diabetes (T2D) GWAS signals and infer underlying regulatory mechanisms. We make a conceptual analogy between natural language and human genome and demonstrate LOGO is an accurate, fast, scalable, and robust framework to interpret non-coding regions for global sequence labeling as well as for variant prioritization at base-resolution.

LOGO, a contextualized pre-trained language model of human genome flexibly adapts to various downstream tasks by fine-tuning

Interpretation of non-coding genome remains an unsolved challenge in human genetics due to impracticality of exhaustively annotate biochemically active elements in all conditions. Deep learning based computational approaches emerge recently to help interpretating non-coding regions. Here we present LOGO (Language of Genome), a self-attention based contextualized pre-trained language model that applies self-supervision techniques to learn bidirectional representations of unlabeled human reference genome and extend to a series of downstream tasks via fine-tuning. We also explore a novel knowledge embedded version of LOGO to incorporate prior human annotations. Experiments show that LOGO achieves 15% absolute improvement for promoter identification and up to 4.5% absolute improvement for enhancer-promoter interaction prediction. LOGO exhibits state-of-the-art predictive power on chromatin features with only 3% parameterization against fully supervised convolutional neural network, DeepSEA. Fine-tuned LOGO also shows outstanding performance in prioritizing non-coding variants associated with human diseases. In addition, we apply LOGO to interpret type 2 diabetes (T2D) GWAS signals and infer underlying regulatory mechanisms. We make a conceptual analogy between natural language and human genome and demonstrate LOGO is an accurate, fast, scalable, and robust framework with powerful adaptability to various tasks without substantial task-specific architecture modifications.

Chariton and Latin Elegy I

Chapter 1 establishes Latin love elegy (especially Propertius) as an important frame of reference for Chariton, and explores a number of characteristics (lexical and thematic) that all constitute an extreme or ‘totalizing’ attitude towards love on the part of the lover. Section 1.2 addresses the language of wholeness and exclusion (ὅλος‎ and μόνος‎; totus and solus) on display in Chariton and elegy, which is suggestive of a direct link. Section 1.3 approaches the conceptual analogy between love and death in Chariton and elegy, and argues that Chariton looks to the Latin poets for his characterization of Chaereas and Dionysius as obsessed with death and erotically motivated thoughts of suicide (especially in connection with the lover who imagines his own funeral). Section 1.4 similarly approaches the characters of Chaereas and Dionysius as susceptible to overwhelming jealousy, the quintessential ‘elegiac passion’; as well as a number of Propertian poems, this section also argues for an extended allusion to Ovid’s treatment of the Procris and Cephalus myth as narrated in Ars 3 and the Metamorphoses. Thematic proximities between Chariton and the Latin poets are supported by strikingly close points of verbal contact.

Militant memocracy in International Relations: Mnemonical status anxiety and memory laws in Eastern Europe

This article theorises the nexus between mnemonical status anxiety and militant memory laws. Extending the understanding of status-seeking in international relations to the realm of historical memory, I argue that the quest for mnemonical recognition is a status struggle in an international social hierarchy of remembering constitutive events of the past. A typology of mnemopolitical status-seeking is presented on the example of Russia (mnemonical positionalism), Poland (mnemonical revisionism), and Ukraine (mnemonical self-emancipation). Memory laws provide a common instance of securing and/or improving a state's mnemonical standing in the relevant memory order. Drawing on the conceptual analogy of militant democracy, the article develops the notion militant memocracy, or the governance of historical memory through a dense network of prescribing and proscribing memory laws and policies. Similar to its militant democracy counterpart, militant memocracy is in danger of self-inflicted harm to the object of defence in the very effort to defend it: its precautionary and punitive measures resound rather than fix the state's mnemonical anxiety problem.

Entangled dynamos and Joule heating in the Earth's ionosphere

The Earth's neutral atmosphere is the driver of the well-known solar quiet (Sq) and other magnetic variations observed for more than 100 years. Yet the understanding of how the neutral wind can accomplish a dynamo effect has been incomplete. A new viable model is presented where a dynamo effect is obtained only in the case of winds perpendicular to the magnetic field B that do not map along B. Winds where u×B is constant have no effect. We identify Sq as being driven by wind differences at magnetically conjugate points and not by a neutral wind per se. The view of two different but entangled dynamos is favoured, with some conceptual analogy to quantum mechanical states. Because of the large preponderance of the neutral gas mass over the ionized component in the Earth's ionosphere, the dominant effect of the plasma adjusting to the winds is Joule heating. The amount of global Joule heating power from Sq is estimated, with uncertainties, to be much lower than Joule heating from ionosphere–magnetosphere coupling at high latitudes in periods of strong geomagnetic activity. However, on average both contributions could be relatively comparable. The global contribution of heating by ionizing solar radiation in the same height range should be 2–3 orders of magnitude larger.

Introduction

Corporations have limited responsibilities in international law but enjoy far-reaching rights and privileges. International legal debates often conceive of this issue as a problem of business accountability for human rights violations. Conceptually, the issue of corporations in international law has focused on whether or not they are, or ought to be, recognized as ‘subjects’ of responsibility in international law and on the adequate conceptual analogy to the corporation. The introduction presents an alternative way of thinking about the role of international law and its relevance to the private business corporation. It traces the emergence of the contemporary legal architecture for corporations in international law and shows how modern international law constitutes a framework within which businesses and governments allocate resources and responsibilities—a framework that began to operate as early as the late-nineteenth century and continued throughout the twentieth century.

(Post)phenomenological Approach to Homo Sapiens Technicus

In this paper I use a (post)phenomenological approach to clarify the objective cultural expansion of our technology. Thus, I establish a conceptual analogy between two different philosophical analyses of human–machine relations – one historical and one phenomenological. I develop the analogy between them and their corresponding concepts in several steps. (1) First, I present the Homo sapiens technicus tendency and then the phenomenological differentiation between body schema and body image. All of these elucidate our involvement with machines. (2) Then, I conceptualize the term ‘context’, coupling its structural stability with the idea of distextaulity in order to achieve a better empirical understanding of our technological contradictions. (3) I continue to develop and enrich the analogy by illuminating the functional similarities – fluid boundary, automation, complexity – between contextual structures on the one hand and body schemata on the other. (4) Finally, I explore a deeper causal and narrative connection between those strands, shedding light on an interesting twofold circularity: a circular causation and a double narrative within Homo sapiens technicus.

Entangled Dynamos and Joule Heating in the Earth’s Ionosphere

The Earth's neutral atmosphere is the driver of the well-known Solar quiet (Sq) and other magnetic variations, observed since more than 100 years. Yet the understanding of how the neutral wind can accomplish a dynamo effect has been incomplete. A new viable model is presented, where a dynamo effect is obtained only in case of winds perpendicular to the magnetic field B that spatially vary along B. Uniform winds have no effect. We identify Sq as being driven by wind differences at magnetically conjugate points, and not by a neutral wind per se. The view of two different but entangled dynamos is favoured, with some conceptual analogy to quantum mechanical states. Because of the large preponderance of the neutral gas mass over the ionized component in the Earth's ionosphere the dominant effect of the plasma adjusting to the winds is Joule heating. The amount of global Joule heating power from Sq is estimated, with uncertainties, to be much lower than Joule heating from ionosphere-magnetosphere coupling at high latitudes in periods of strong geomagnetic activity. However, on average both contributions could be relatively evenly matched. The global contribution of heating by ionizing solar radiation in the same height range should be 2–3 orders of magnitude larger.

The Speed of Light

You have probably heard that nothing can go faster than the speed of light. In this chapter, we will see that the speed of light is even more remarkable: it is the same in all frames. From this surprising fact, we will deduce that this speed must also serve as a limit and that Galilean velocity addition fails to describe how nature works at high speeds: our first glimpses of the modern understanding of relativity. To make you more comfortable with the fact that velocities must add nonlinearly, this chapter develops an intuitive conceptual analogy for velocity addition.