Motivation: Some first order methods for protein sequence analysis inherently treat each position as independent. We develop a general framework for introducing longer range interactions. We then demonstrate the power of our approach by applying it to secondary structure prediction; under the independence assumption, sequences produced by existing methods can produce features that are not protein like, an extreme example being a helix of length 1. Our goal was to make the predictions from state of the art methods more realistic, without loss of performance by other measures.Results: Our framework for longer range interactions is described as a k-mer order model. We succeeded in applying our model to the specific problem of secondary structure prediction, to be used as an additional layer on top of existing methods. We achieved our goal of making the predictions more realistic and protein like, and remarkably this also improved the overall performance. We improve the Segment OVerlap (SOV) score by 1.8%, but more importantly we radically improve the probability of the real sequence given a prediction from an average of 0.271 per residue to 0.385. Crucially, this improvement is obtained using no additional information.Availability: http://supfam.cs.bris.ac.uk/kmerContact: gough@cs.bris.ac.uk

Understanding how the brain forms representations of structured information distributed in time is a challenging endeavour for the neuroscientific community, requiring computationally and neurobiologically informed approaches. The neural mechanisms for segmenting continuous streams of sensory input and establishing representations of dependencies remain largely unknown, as do the transformations and computations occurring between the brain regions involved in these aspects of sequence processing. We propose a blueprint for a neurobiologically informed and informing computational model of sequence processing (entitled: Vector-symbolic Sequencing of Binding INstantiating Dependencies, or VS-BIND). This model is designed to support the transformation of serially ordered elements in sensory sequences into structured representations of bound dependencies, readily operates on multiple timescales, and encodes or decodes sequences with respect to chunked items wherever dependencies occur in time. The model integrates established vector symbolic additive and conjunctive binding operators with neurobiologically plausible oscillatory dynamics, and is compatible with modern spiking neural network simulation methods. We show that the model is capable of simulating previous findings from structured sequence processing tasks that engage fronto-temporal regions, specifying mechanistic roles for regions such as prefrontal areas 44/45 and the frontal operculum during interactions with sensory representations in temporal cortex. Finally, we are able to make predictions based on the configuration of the model alone that underscore the importance of serial position information, which requires input from time-sensitive cells, known to reside in the hippocampus and dorsolateral prefrontal cortex. This article is part of the theme issue ‘Towards mechanistic models of meaning composition’.

Background: Animal models of stroke have been criticised as having poor predictive validity, lacking risk factors prevalent in an aging population. This pilot study examined the development of comorbidities in a combined aged and high-fat diet model, and then examined the feasibility of modelling stroke in such rats. Methods: Twelve-month old male Wistar-Han rats (n=15) were fed a 60% fat diet for 8 months during which monthly serial blood samples were taken to assess the development of metabolic syndrome and pro-inflammatory markers. Following this, to pilot the suitability of these rats for undergoing surgical models of stroke, they underwent 30min of middle cerebral artery occlusion (MCAO) alongside younger controls fed a standard diet (n=10). Survival, weight and functional outcome were monitored, and blood vessels and tissues collected for analysis. Results: A high fat diet in aged rats led to substantial obesity. These rats did not develop type 2 diabetes or hypertension. There was thickening of the thoracic arterial wall and vacuole formation in the liver; but of the cytokines examined changes were not seen. MCAO surgery and behavioural assessment was possible in this model (with some caveats discussed in manuscript). Conclusions: This study shows MCAO is possible in aged, obese rats. However, this model is not ideal for recapitulating the complex comorbidities commonly seen in stroke patients.

The human brain extracts meaning from the world using an extensive neural system for semantic knowledge. Whether such broadly distributed systems crucially depend on or can compensate for the loss of one of their highly interconnected hubs is controversial. The strongest level of causal evidence for the role of a brain hub is to evaluate its acute network-level impact following disconnection and any rapid functional compensation that ensues. We report rare neurophysiological data from two patients who underwent awake intracranial recordings during a speech prediction task immediately before and after neurosurgical treatment that required disconnection of the anterior temporal lobe (ATL), a crucial hub for semantic knowledge. Informed by a predictive coding framework, we tested three sets of hypotheses, including diaschisis causing disruption in interconnected sites and incomplete or complete compensation by other language-critical and speech processing sites. Immediately after ATL disconnection, we observed highly specific neurophysiological alterations in the recorded fronto-temporal network, including abnormally magnified high gamma responses to the speech sounds in auditory cortex. We also observed evidence for rapid compensation, seen as focal increases in effective connectivity involving language-critical sites in the inferior frontal gyrus and speech processing sites in auditory cortex. However, compensation was incomplete, in part because after ATL disconnection speech prediction signals were depleted in auditory cortex. This study provides direct causal evidence for a semantic hub in the human brain and shows striking neural impact and a rapid attempt at compensation in a neural network after the loss of one of its hubs.