Predicting Binding from Screening Assays with Transformer Network Embeddings.

Cheminformatics aims to assist in chemistry applications that depend on molecular interactions, structural characteristics, and functional properties. The arrival of deep learning and the abundance of easily accessible chemical data from repositories like PubChem have enabled advancements in computer-aided drug discovery. Virtual high-throughput screening (vHTS) is one such technique that integrates chemical domain knowledge to perform in silico biomolecular simulations, but prediction of binding affinity is restricted due to limited availability of ground-truth binding assay results. Here, text representations of 83 000 000 molecules are leveraged to perform single-target binding affinity prediction directly on the outcome of screening assays. The embedding of an end-to-end transformer neural network, trained to encode the structural characteristics of a molecule via a text-based translation task, is repurposed through transfer learning to classify binding affinity to single targets with few known binding compounds. We quantify the observed increase in AUC on binding prediction tasks between classifiers trained on the translation embedding versus those using an untrained embedding. Visualization of the embedding space reveals organization of structural and functional properties that aid binding prediction. The pretrained transformer, data, and associated software to extract embeddings are made publicly available at https://github.com/mpcrlab/MolecularTransformerEmbeddings.

Intrinsic and contextual features in object recognition.

The context in which an object is found can facilitate its recognition. Yet, it is not known how effective this contextual information is relative to the object's intrinsic visual features, such as color and shape. To address this, we performed four experiments using rendered scenes with novel objects. In each experiment, participants first performed a visual search task, searching for a uniquely shaped target object whose color and location within the scene was experimentally manipulated. We then tested participants' tendency to use their knowledge of the location and color information in an identification task when the objects' images were degraded due to blurring, thus eliminating the shape information. In Experiment 1, we found that, in the absence of any diagnostic intrinsic features, participants identified objects based purely on their locations within the scene. In Experiment 2, we found that participants combined an intrinsic feature, color, with contextual location in order to uniquely specify an object. In Experiment 3, we found that when an object's color and location information were in conflict, participants identified the object using both sources of information equally. Finally, in Experiment 4, we found that participants used whichever source of information-either color or location-was more statistically reliable in order to identify the target object. Overall, these experiments show that the context in which objects are found can play as important a role as intrinsic features in identifying the objects.

Visual fixations during processing of time-compressed audiovisual presentations.

Time-compression is a technique that allows users to adjust the playback speed of audio recordings, but comprehension declines at higher speeds. Previous research has shown that under challenging auditory conditions people have a greater tendency to fixate regions closer to a speaker's mouth. In the current study, we investigated whether there is a similar tendency to fixate the mouth region for time-compressed stimuli. Participants were presented with a brief audiovisual lecture at different speeds, while eye fixations were recorded, and comprehension was tested. Results showed that the 50% compressed lecture group looked more at the nose compared to eye fixations for the normal lecture, and those in the 75% compressed group looked more towards the mouth. Greater compression decreased comprehension, but audiovisual information did not reduce this deficit. These results indicate that people seek out audiovisual information to overcome time-compression, demonstrating the flexibility of the multimodal attentional system.

An Imagination - Procrastination Link? The Role of Efficacy Beliefs, Visual Imagery, and Affect in Academic Procrastination.

Previous studies have established that there is a relationship between efficacy beliefs and procrastination. Theory and research on motivation suggest that visual imagery (the capacity to create vivid mental images) may be implicated in this relationship and in the general tendency to procrastinate. This study's aim was to build on prior work by examining the role of visual imagery, as well as roles of other specific personal and affective factors, in predicting academic procrastination. Self-efficacy for self regulatory behavior was observed to be the strongest predictor, predicting lower rates of academic procrastination, though this effect was significantly greater for individuals who scored higher on a measure of visual imagery. Visual imagery predicted higher levels of academic procrastination when included in a regression model with other significant factors, though this relationship did not hold for individuals who scored higher on self regulatory self-efficacy, suggesting that this self-belief may shield individuals who would otherwise be disposed to procrastination behavior. Negative affect was observed to predict higher levels of academic procrastination, contrary to a previous finding. This result highlightsthe importance of considering social contextual issues that may influence emotional states, such as those surrounding the Covid-19 epidemic, in studies of procrastination.

Predicting residues involved in anti-DNA autoantibodies with limited neural networks.

Computer-aided rational vaccine design (RVD) and synthetic pharmacology are rapidly developing fields that leverage existing datasets for developing compounds of interest. Computational proteomics utilizes algorithms and models to probe proteins for functional prediction. A potentially strong target for computational approach is autoimmune antibodies, which are the result of broken tolerance in the immune system where it cannot distinguish "self" from "non-self" resulting in attack of its own structures (proteins and DNA, mainly). The information on structure, function, and pathogenicity of autoantibodies may assist in engineering RVD against autoimmune diseases. Current computational approaches exploit large datasets curated with extensive domain knowledge, most of which include the need for many resources and have been applied indirectly to problems of interest for DNA, RNA, and monomer protein binding. We present a novel method for discovering potential binding sites. We employed long short-term memory (LSTM) models trained on FASTA primary sequences to predict protein binding in DNA-binding hydrolytic antibodies (abzymes). We also employed CNN models applied to the same dataset for comparison with LSTM. While the CNN model outperformed the LSTM on the primary task of binding prediction, analysis of internal model representations of both models showed that the LSTM models recovered sub-sequences that were strongly correlated with sites known to be involved in binding. These results demonstrate that analysis of internal processes of LSTM models may serve as a powerful tool for primary sequence analysis.

Mice recognize 3D objects from recalled 2D pictures, support for picture-object equivalence.

Picture-object equivalence or recognizing a three-dimensional (3D) object after viewing a two-dimensional (2D) photograph of that object, is a higher-order form of visual cognition that may be beyond the perceptual ability of rodents. Behavioral and neurobiological mechanisms supporting picture-object equivalence are not well understood. We used a modified visual recognition memory task, reminiscent of those used for primates, to test whether picture-object equivalence extends to mice. Mice explored photographs of an object during a sample session, and 24 h later were presented with the actual 3D object from the photograph and a novel 3D object, or the stimuli were once again presented in 2D form. Mice preferentially explored the novel stimulus, indicating recognition of the "familiar" stimulus, regardless of whether the sample photographs depicted radially symmetric or asymmetric, similar, rotated, or abstract objects. Discrimination did not appear to be guided by individual object features or low-level visual stimuli. Inhibition of CA1 neuronal activity in dorsal hippocampus impaired discrimination, reflecting impaired memory of the 2D sample object. Collectively, results from a series of experiments provide strong evidence that picture-object equivalence extends to mice and is hippocampus-dependent, offering important support for the appropriateness of mice for investigating mechanisms of human cognition.

Screening for Early-Stage Alzheimer's Disease Using Optimized Feature Sets and Machine Learning.

BACKGROUND: Detecting early-stage Alzheimer's disease in clinical practice is difficult due to a lack of efficient and easily administered cognitive assessments that are sensitive to very mild impairment, a likely contributor to the high rate of undetected dementia. OBJECTIVE: We aim to identify groups of cognitive assessment features optimized for detecting mild impairment that may be used to improve routine screening. We also compare the efficacy of classifying impairment using either a two-class (impaired versus non-impaired) or three-class using the Clinical Dementia Rating (CDR 0 versus CDR 0.5 versus CDR 1) approach. METHODS: Supervised feature selection methods generated groups of cognitive measurements targeting impairment defined at CDR 0.5 and above. Random forest classifiers then generated predictions of impairment for each group using highly stochastic cross-validation, with group outputs examined using general linear models. RESULTS: The strategy of combining impairment levels for two-class classification resulted in significantly higher sensitivities and negative predictive values, two metrics useful in clinical screening, compared to the three-class approach. Four features (delayed WAIS Logical Memory, trail-making, patient and informant memory questions), totaling about 15 minutes of testing time (∼30 minutes with delay), enabled classification sensitivity of 94.53% (88.43% positive predictive value, PPV). The addition of four more features significantly increased sensitivity to 95.18% (88.77% PPV) when added to the model as a second classifier. CONCLUSION: The high detection rate paired with the minimal assessment time of the four identified features may act as an effective starting point for developing screening protocols targeting cognitive impairment defined at CDR 0.5 and above.

Online surveillance of novel psychoactive substances (NPS): Monitoring Reddit discussions as a predictor of increased NPS-related exposures.

BACKGROUND: Novel psychoactive substances (NPS) present continuous and growing challenges for the scientific, medical, and interventional communities as emerging substances on recreational drug markets change national and international drug landscapes. NPS account for an increasing proportion of adverse events, hospitalizations, and deaths due to increasing potency and unanticipated biological effects compared to predecessors. This study evaluated the utility of drug use forums as an early indicator or predictor of impending intoxications with potentially harmful or lethal outcomes prior to their occurrences. METHODS: Eight NPS were selected for evaluation to assess the relationship between online mentions of drugs and their involvement in toxic exposures or overdoses. Mentions on Reddit drug forum discussions were tallied and toxicology testing results from forensic investigations in the US were assessed. The selected NPS covered several subclasses and a predetermined time range (2013-2020). They included carfentanil, U-47700, eutylone, flualprazolam, N-ethylpentylone, 5F-MDMB-PICA, isotonitazene, and brorphine. RESULTS: Seven NPS (excluding 5F-MDMB-PICA) appeared in discussions on Reddit prior to their implication in poisonings or intoxications. Distinct increases and decreases in number of mentions and number of exposures were observed. For most substances (n = 5, 63%), a rise in Reddit mentions was soon followed by a corresponding rise in toxicology positivity. Peak positivity for carfentanil and flualprazolam, however, preceded peak Reddit mentions. CONCLUSIONS: This study demonstrated the utility of social media sites, such as Reddit, as a predictor for future trends in NPS-related exposures. These results provide confirmation that activity on drug use forums in the virtual world can help predict changes in exposures associated with new or re-emerging NPS in the real world. The results warrant further evaluation as a strategy for inclusion in early warning systems.

Convexities move because they contain matter.

Figure-ground assignment to a contour is a fundamental stage in visual processing. The current paper introduces a novel, highly general dynamic cue to figure-ground assignment: "Convex Motion." Across six experiments, subjects showed a strong preference to assign figure and ground to a dynamically deforming contour such that the moving contour segment was convex rather than concave. Experiments 1 and 2 established the preference across two different kinds of deformational motion. Additional experiments determined that this preference was not due to fixation (Experiment 3) or attentional mechanisms (Experiment 4). Experiment 5 found a similar, but reduced bias for rigid-as opposed to deformational-motion, and Experiment 6 demonstrated that the phenomenon depends on the global motion of the effected contour. An explanation of this phenomenon is presented on the basis of typical natural deformational motion, which tends to involve convex contour projections that contain regions consisting of physical "matter," as opposed to concave contour indentations that contain empty space. These results highlight the fundamental relationship between figure and ground, perceived shape, and the inferred physical properties of an object.

Perception of being observed by a speaker alters gaze behavior.

Previous research has shown that gaze behavior of a speaker's face during speech encoding is influenced by an array of factors relating to the quality of the speech signal and the encoding task. In these studies, participants were aware they were viewing pre-recorded stimuli of a speaker that is not representative of natural social interactions in which an interlocutor can observe one's gaze direction, potentially affecting fixation behavior due to communicative and social considerations. To assess the potential role of these factors during speech encoding, we compared fixation behavior during a speech-encoding task under two conditions: in the "real-time" condition, we used deception to convince participants that they were interacting with a live person who was able to see and hear them through online remote video communication. In the "pre-recorded" condition, participants were correctly informed they were watching a previously recorded video. We found that participants fixated the interlocutor's face significantly less in the real-time condition than the pre-recorded condition. When participants did look at the face, they fixated the mouth at a higher proportion of the time in the pre-recorded condition versus the real-time condition. These findings suggest that people engage in avoidance of potentially useful speech-directed fixations when they believe their fixations are being observed and demonstrate that social factors play a significant role in fixation behavior during speech encoding.

Machine-learning approaches to substance-abuse research: emerging trends and their implications.

PURPOSE OF REVIEW: To provide an accessible overview of some of the most recent trends in the application of machine learning to the field of substance use disorders and their implications for future research and practice. RECENT FINDINGS: Machine-learning (ML) techniques have recently been applied to substance use disorder (SUD) data for multiple predictive applications including detecting current abuse, assessing future risk and predicting treatment success. These models cover a wide range of machine-learning techniques and data types including physiological measures, longitudinal surveys, treatment outcomes, national surveys, medical records and social media. SUMMARY: The application of machine-learning models to substance use disorder data shows significant promise, with some use cases and data types showing high predictive accuracy, particularly for models of physiological and behavioral measures for predicting current substance use, portending potential clinical diagnostic applications; however, these results are uneven, with some models performing poorly or at chance, a limitation likely reflecting insufficient data and/or weak validation methods. The field will likely benefit from larger and more multimodal datasets, greater standardization of data recording and rigorous testing protocols as well as greater use of modern deep neural network models applied to multimodal unstructured datasets.

Gender Perception From Gait: A Comparison Between Biological, Biomimetic and Non-biomimetic Learning Paradigms.

This paper explores in parallel the underlying mechanisms in human perception of biological motion and the best approaches for automatic classification of gait. The experiments tested three different learning paradigms, namely, biological, biomimetic, and non-biomimetic models for gender identification from human gait. Psychophysical experiments with twenty-one observers were conducted along with computational experiments without applying any gender specific modifications to the models or the stimuli. Results demonstrate the utilization of a generic memory based learning system in humans for gait perception, thus reducing ambiguity between two opposing learning systems proposed for biological motion perception. Results also support the biomimetic nature of memory based artificial neural networks (ANN) in their ability to emulate biological neural networks, as opposed to non-biomimetic models. In addition, the comparison between biological and computational learning approaches establishes a memory based biomimetic model as the best candidate for a generic artificial gait classifier (83% accuracy, p < 0.001), compared to human observers (66%, p < 0.005) or non-biomimetic models (83%, p < 0.001) while adhering to human-like sensitivity to gender identification, promising potential for application of the model in any given non-gender based gait perception objective with superhuman performance.

Figure-ground assignment to a translating contour: a preference for advancing vs. receding motion.

Past research on figure-ground assignment to contours has largely considered static stimuli. Here we report a simple and extremely robust dynamic cue to figural assignment, based on whether the bounding region of a contour is growing larger within the field of view ("advancing") rather than smaller ("receding"). Subjects viewed a straight or jagged contour dividing two colored regions translating behind a virtual aperture and had to report which color they had seen "moving in front", effectively assigning figure to that side of the contour. Across three experiments, subjects showed a strong preference to assign figure such that the bounded contour was advancing. This was true regardless of the direction of motion of the contour and regardless of the initial/ending size of the bounded regions (i.e., the motion cue served to override the conventional cue to figure-ground of smaller area). In a fourth, control experiment, subjects showed no such bias when it was the aperture, rather than the contour, that moved, demonstrating that the effect depends on contour motion and not simply an increase in area. We discuss a possible explanation for this bias as well as the general implications regarding dynamic factors in form perception.

Visual judgment of similarity across shape transformations: evidence for a compositional model of articulated objects.

A single biological object, such as a hand, can assume multiple, very different shapes, due to the articulation of its parts. Yet we are able to recognize all of these shapes as examples of the same object. How is this invariance to pose achieved? Here, we present evidence that the visual system maintains a model of object transformation that is based on rigid, convex parts articulating at extrema of negative curvature, i.e., part boundaries. We compared similarity judgments in a task in which subjects had to decide which of the two transformed versions of a 'base' shape-one a 'biologically valid' articulation and one a geometrically similar but 'biologically invalid' articulation-was more similar to the base shape. Two types of comparisons were made: in the figure/ground-reversal, the invalid articulation consisted of exactly the same contour transformation as the valid one with reversed figural polarity. In the axis-of-rotation reversal, the valid articulation consisted of a part rotated around its concave part boundaries, while the invalid articulation consisted of the same part rotated around the endpoints on the opposite side of the part. In two separate 2AFC similarity experiments-one in which the base and transformed shapes were presented simultaneously and one in which they were presented sequentially-subjects were more likely to match the base shape to a transform when it corresponded to a legitimate articulation. These results suggest that the visual system maintains expectations about the way objects will transform, based on their static geometry.

Determination of visual figure and ground in dynamically deforming shapes.

Figure/ground assignment - determining which part of the visual image is foreground and which background - is a critical step in early visual analysis, upon which much later processing depends. Previous research on the assignment of figure and ground to opposing sides of a contour has almost exclusively involved static geometric factors - such as convexity, symmetry, and size - in non-moving images. Here, we introduce a new class of cue to figural assignment based on the motion of dynamically deforming contours. Subjects viewing an animated, deforming shape tended to assign figure and ground so that articulating curvature extrema - i.e., "hinging" vertices - had negative (concave) contour curvature. This articulating-concavity bias is present when all known static cues to figure/ground are absent or neutral in each of the individual frames of the animation, and even seems to override a number of well-known static cues when they are in opposition to the motion cue. We propose that the phenomenon reflects the visual system's inbuilt expectations about the way shapes will deform - specifically, that deformations tend to involve rigid parts articulating at concavities.

Language familiarity modulates relative attention to the eyes and mouth of a talker.

We investigated whether the audiovisual speech cues available in a talker's mouth elicit greater attention when adults have to process speech in an unfamiliar language vs. a familiar language. Participants performed a speech-encoding task while watching and listening to videos of a talker in a familiar language (English) or an unfamiliar language (Spanish or Icelandic). Attention to the mouth increased in monolingual subjects in response to an unfamiliar language condition but did not in bilingual subjects when the task required speech processing. In the absence of an explicit speech-processing task, subjects attended equally to the eyes and mouth in response to both familiar and unfamiliar languages. Overall, these results demonstrate that language familiarity modulates selective attention to the redundant audiovisual speech cues in a talker's mouth in adults. When our findings are considered together with similar findings from infants, they suggest that this attentional strategy emerges very early in life.

What change detection tells us about the visual representation of shape.

Many recent findings suggest that human observers are surprisingly "blind" to changes in visual displays, failing to notice when substantial scene elements are added, subtracted, or altered in successive presentations of the scene. But observers are far more sensitive to certain visual changes than others, and we suggest that which types of changes enjoy differential sensitivity can reveal a great deal about the underlying visual representations. In this study, we investigate how the human visual system represents the shape of objects by demonstrating a previously unknown influence on detection of changes in shape: the sign of contour curvature. We show that subjects are substantially more sensitive to changes in concave regions of a shape's contour than to changes in convex regions, even when these changes do not alter the number or location of parts. Further, we show that this effect is modulated by figure-ground assignment, so that changes to the same physical contour are more or less detectable, depending on the contour's perceived figural status, which determines whether the change falls in a concave or convex region. The results demonstrate a heightened sensitivity for changes at concavities that is not reducible to a sensitivity to changes in gross part structure.

Detection of change in shape: an advantage for concavities.

Shape representation was studied using a change detection task. Observers viewed two individual shapes in succession, either identical or one a slightly altered version of the other, and reported whether they detected a change. We found a dramatic advantage for concave compared to convex changes of equal magnitude. Observers were more accurate when a concavity along the contour was introduced, or removed, compared to a convexity. This result sheds light on the underlying representation of visual shape, and in particular the central role played by part-boundaries. Moreover, this finding shows how change detection methodology can serve as a useful tool in studying the specific form of visual representations.