Search

Bacterial small RNAs (sRNAs) are important regulators of gene expression; however, the impact of natural mutations on sRNA functions has not been studied extensively. Here we show that the sRNA MgrR contains a unique 53 bp insertion in Escherichia fergusonii , a close relative of Escherichia coli and Salmonella enterica . The insertion is a repetitive extragenic palindromic (REP) sequence that could block transcription, but full-length MgrR is produced in E. fergusonii , showing that the insertion has not affected sRNA production. Additionally, despite containing the large insertion, the sRNA appears to be functional because deletion of mgrR made E. fergusonii more susceptible to H2O2. The molecular details of MgrR’s roles in H2O2defence are yet to be defined, but our results suggest that having an alternative function allowed the sRNA to be retained in E. fergusonii despite it sustaining a large, potentially disruptive mutation.

Existing research on making sense of deep neural networks often focuses on neuron-level interpretation, which may not adequately capture the bigger picture of how concepts are collectively encoded by multiple neurons. We present NeuroCartography, an interactive system that scalably summarizes and visualizes concepts learned by neural networks. It automatically discovers and groups neurons that detect the same concepts, and describes how such neuron groups interact to form higher-level concepts and the subsequent predictions. NeuroCartography introduces two scalable summarization techniques: (1) neuron clustering groups neurons based on the semantic similarity of the concepts detected by neurons (e.g., neurons detecting "dog faces" of different breeds are grouped); and (2) neuron embedding encodes the associations between related concepts based on how often they co-occur (e.g., neurons detecting "dog face" and "dog tail" are placed closer in the embedding space). Key to our scalable techniques is the ability to efficiently compute all neuron pairs' relationships, in time linear to the number of neurons instead of quadratic time. NeuroCartography scales to large data, such as the ImageNet dataset with 1.2M images. The system's tightly coordinated views integrate the scalable techniques to visualize the concepts and their relationships, projecting the concept associations to a 2D space in Neuron Projection View, and summarizing neuron clusters and their relationships in Graph View. Through a large-scale human evaluation, we demonstrate that our technique discovers neuron groups that represent coherent, human-meaningful concepts. And through usage scenarios, we describe how our approaches enable interesting and surprising discoveries, such as concept cascades of related and isolated concepts. The NeuroCartography visualization runs in modern browsers and is open-sourced., Comment: Accepted to IEEE VIS'21

Interferon-lambda (IFN-λ) protects intestinal epithelial cells (IECs) from enteric viruses by inducing expression of antiviral IFN-stimulated genes (ISGs). Here, we find that bacterial microbiota stimulate a homeostatic ISG signature in the intestine of specific pathogen-free mice. This homeostatic ISG expression is restricted to IECs, depends on IEC-intrinsic expression of IFN-λ receptor (Ifnlr1), and is associated with IFN-λ production by leukocytes. Strikingly, imaging of these homeostatic ISGs reveals localization to pockets of the epithelium and concentration in mature IECs. Correspondingly, a minority of mature IECs express these ISGs in public single-cell RNA sequencing datasets from mice and humans. Furthermore, we assessed the ability of orally-administered bacterial components to restore localized ISGs in mice lacking bacterial microbiota. Lastly, we find that IECs lacking Ifnlr1 are hyper-susceptible to initiation of murine rotavirus infection. These observations indicate that bacterial microbiota stimulate ISGs in localized regions of the intestinal epithelium at homeostasis, thereby preemptively activating antiviral defenses in vulnerable IECs to improve host defense against enteric viruses.

With the widespread use of toxic language online, platforms are increasingly using automated systems that leverage advances in natural language processing to automatically flag and remove toxic comments. However, most automated systems -- when detecting and moderating toxic language -- do not provide feedback to their users, let alone provide an avenue of recourse for these users to make actionable changes. We present our work, RECAST, an interactive, open-sourced web tool for visualizing these models' toxic predictions, while providing alternative suggestions for flagged toxic language. Our work also provides users with a new path of recourse when using these automated moderation tools. RECAST highlights text responsible for classifying toxicity, and allows users to interactively substitute potentially toxic phrases with neutral alternatives. We examined the effect of RECAST via two large-scale user evaluations, and found that RECAST was highly effective at helping users reduce toxicity as detected through the model. Users also gained a stronger understanding of the underlying toxicity criterion used by black-box models, enabling transparency and recourse. In addition, we found that when users focus on optimizing language for these models instead of their own judgement (which is the implied incentive and goal of deploying automated models), these models cease to be effective classifiers of toxicity compared to human annotations. This opens a discussion for how toxicity detection models work and should work, and their effect on the future of online discourse., Comment: 26 pages, 5 figures, CSCW '21

As toxic language becomes nearly pervasive online, there has been increasing interest in leveraging the advancements in natural language processing (NLP), from very large transformer models to automatically detecting and removing toxic comments. Despite the fairness concerns, lack of adversarial robustness, and limited prediction explainability for deep learning systems, there is currently little work for auditing these systems and understanding how they work for both developers and users. We present our ongoing work, RECAST, an interactive tool for examining toxicity detection models by visualizing explanations for predictions and providing alternative wordings for detected toxic speech., Comment: 8 Pages, 3 figures, The eighth International Workshop of Chinese CHI Proceedings

Substances involved in overdose deaths have shifted over time and continue to undergo transition. Early detection of emerging drugs involved in overdose is a major challenge for traditional public health data systems. While novel social media data have shown promise, there is a continued need for robust natural language processing approaches that can identify emerging substances. Consequently, we developed a new metric, the relative similarity ratio, based on diachronic word embeddings to measure movement in the semantic proximity of individual substance words to 'overdose' over time. Our analysis of 64,420,376 drug-related posts made between January 2011 and December 2018 on Reddit, the largest online forum site, reveals that this approach successfully identified fentanyl, the most significant emerging substance in the overdose epidemic, >1 year earlier than traditional public health data systems. Use of diachronic word embeddings may enable improved identification of emerging substances involved in drug overdose, thereby improving the timeliness of prevention and treatment activities.

States facing the decision to develop a nuclear weapons program do so within a broader context of their relationships with other countries. How these diplomatic, economic, and strategic relationships impact proliferation decisions, however, remains under-specified. Adding to the existing empirical literature that attempts to model state proliferation decisions, this article introduces the first quantitative heterogeneous network analysis of how networks of conflict, alliances, trade, and nuclear cooperation interact to spur or deter nuclear proliferation. Using a multiplex network model, we conceptualize states as nodes linked by different modes of interaction represented on individual network layers. Node strength is used to quantify factors correlated with nuclear proliferation and these are combined in a weighted sum across layers to provide a metric characterizing the proliferation behavior of the state. This multiplex network modeling approach provides a means for identifying states with the highest relative likelihood of proliferation—based only on their relationships to other states. This work demonstrates that latent conflict and nuclear cooperation are positively correlated with proliferation, while an increased trade dependence suggests a decreased proliferation likelihood. A case study on Iran’s controversial nuclear program and past nuclear activity is also provided. These findings have clear, policy-relevant conclusions related to alliance posture, sanctions policy, and nuclear assistance.