Computational validation of clonal and subclonal copy number alterations from bulk tumor sequencing using CNAqc.

Copy number alterations (CNAs) are among the most important genetic events in cancer, but their detection from sequencing data is challenging because of unknown sample purity, tumor ploidy, and general intra-tumor heterogeneity. Here, we present CNAqc, an evolution-inspired method to perform the computational validation of clonal and subclonal CNAs detected from bulk DNA sequencing. CNAqc is validated using single-cell data and simulations, is applied to over 4000 TCGA and PCAWG samples, and is incorporated into the validation process for the clinically accredited bioinformatics pipeline at Genomics England. CNAqc is designed to support automated quality control procedures for tumor somatic data validation.

A Bayesian method to infer copy number clones from single-cell RNA and ATAC sequencing.

Single-cell RNA and ATAC sequencing technologies enable the examination of gene expression and chromatin accessibility in individual cells, providing insights into cellular phenotypes. In cancer research, it is important to consistently analyze these states within an evolutionary context on genetic clones. Here we present CONGAS+, a Bayesian model to map single-cell RNA and ATAC profiles onto the latent space of copy number clones. CONGAS+ clusters cells into tumour subclones with similar ploidy, rendering straightforward to compare their expression and chromatin profiles. The framework, implemented on GPU and tested on real and simulated data, scales to analyse seamlessly thousands of cells, demonstrating better performance than single-molecule models, and supporting new multi-omics assays. In prostate cancer, lymphoma and basal cell carcinoma, CONGAS+ successfully identifies complex subclonal architectures while providing a coherent mapping between ATAC and RNA, facilitating the study of genotype-phenotype maps and their connection to genomic instability.

Robust deep learning object recognition models rely on low frequency information in natural images.

Machine learning models have difficulty generalizing to data outside of the distribution they were trained on. In particular, vision models are usually vulnerable to adversarial attacks or common corruptions, to which the human visual system is robust. Recent studies have found that regularizing machine learning models to favor brain-like representations can improve model robustness, but it is unclear why. We hypothesize that the increased model robustness is partly due to the low spatial frequency preference inherited from the neural representation. We tested this simple hypothesis with several frequency-oriented analyses, including the design and use of hybrid images to probe model frequency sensitivity directly. We also examined many other publicly available robust models that were trained on adversarial images or with data augmentation, and found that all these robust models showed a greater preference to low spatial frequency information. We show that preprocessing by blurring can serve as a defense mechanism against both adversarial attacks and common corruptions, further confirming our hypothesis and demonstrating the utility of low spatial frequency information in robust object recognition.

Understanding Robustness and Generalization of Artificial Neural Networks Through Fourier Masks.

Despite the enormous success of artificial neural networks (ANNs) in many disciplines, the characterization of their computations and the origin of key properties such as generalization and robustness remain open questions. Recent literature suggests that robust networks with good generalization properties tend to be biased toward processing low frequencies in images. To explore the frequency bias hypothesis further, we develop an algorithm that allows us to learn modulatory masks highlighting the essential input frequencies needed for preserving a trained network's performance. We achieve this by imposing invariance in the loss with respect to such modulations in the input frequencies. We first use our method to test the low-frequency preference hypothesis of adversarially trained or data-augmented networks. Our results suggest that adversarially robust networks indeed exhibit a low-frequency bias but we find this bias is also dependent on directions in frequency space. However, this is not necessarily true for other types of data augmentation. Our results also indicate that the essential frequencies in question are effectively the ones used to achieve generalization in the first place. Surprisingly, images seen through these modulatory masks are not recognizable and resemble texture-like patterns.

Machine learning algorithms on eye tracking trajectories to classify patients with spatial neglect.

BACKGROUND AND OBJECTIVE: Eye-movement trajectories are rich behavioral data, providing a window on how the brain processes information. We address the challenge of characterizing signs of visuo-spatial neglect from saccadic eye trajectories recorded in brain-damaged patients with spatial neglect as well as in healthy controls during a visual search task. METHODS: We establish a standardized pre-processing pipeline adaptable to other task-based eye-tracker measurements. We use traditional machine learning algorithms together with deep convolutional networks (both 1D and 2D) to automatically analyze eye trajectories. RESULTS: Our top-performing machine learning models classified neglect patients vs. healthy individuals with an Area Under the ROC curve (AUC) ranging from 0.83 to 0.86. Moreover, the 1D convolutional neural network scores correlated with the degree of severity of neglect behavior as estimated with standardized paper-and-pencil tests and with the integrity of white matter tracts measured from Diffusion Tensor Imaging (DTI). Interestingly, the latter showed a clear correlation with the third branch of the superior longitudinal fasciculus (SLF), especially damaged in neglect. CONCLUSIONS: The study introduces new methods for both the pre-processing and the classification of eye-movement trajectories in patients with neglect syndrome. The proposed methods can likely be applied to other types of neurological diseases opening the possibility of new computer-aided, precise, sensitive and non-invasive diagnostic tools.

Shallow Univariate ReLU Networks as Splines: Initialization, Loss Surface, Hessian, and Gradient Flow Dynamics.

Understanding the learning dynamics and inductive bias of neural networks (NNs) is hindered by the opacity of the relationship between NN parameters and the function represented. Partially, this is due to symmetries inherent within the NN parameterization, allowing multiple different parameter settings to result in an identical output function, resulting in both an unclear relationship and redundant degrees of freedom. The NN parameterization is invariant under two symmetries: permutation of the neurons and a continuous family of transformations of the scale of weight and bias parameters. We propose taking a quotient with respect to the second symmetry group and reparametrizing ReLU NNs as continuous piecewise linear splines. Using this spline lens, we study learning dynamics in shallow univariate ReLU NNs, finding unexpected insights and explanations for several perplexing phenomena. We develop a surprisingly simple and transparent view of the structure of the loss surface, including its critical and fixed points, Hessian, and Hessian spectrum. We also show that standard weight initializations yield very flat initial functions, and that this flatness, together with overparametrization and the initial weight scale, is responsible for the strength and type of implicit regularization, consistent with previous work. Our implicit regularization results are complementary to recent work, showing that initialization scale critically controls implicit regularization via a kernel-based argument. Overall, removing the weight scale symmetry enables us to prove these results more simply and enables us to prove new results and gain new insights while offering a far more transparent and intuitive picture. Looking forward, our quotiented spline-based approach will extend naturally to the multivariate and deep settings, and alongside the kernel-based view, we believe it will play a foundational role in efforts to understand neural networks. Videos of learning dynamics using a spline-based visualization are available at http://shorturl.at/tFWZ2.

Neurally plausible mechanisms for learning selective and invariant representations.

Coding for visual stimuli in the ventral stream is known to be invariant to object identity preserving nuisance transformations. Indeed, much recent theoretical and experimental work suggests that the main challenge for the visual cortex is to build up such nuisance invariant representations. Recently, artificial convolutional networks have succeeded in both learning such invariant properties and, surprisingly, predicting cortical responses in macaque and mouse visual cortex with unprecedented accuracy. However, some of the key ingredients that enable such success-supervised learning and the backpropagation algorithm-are neurally implausible. This makes it difficult to relate advances in understanding convolutional networks to the brain. In contrast, many of the existing neurally plausible theories of invariant representations in the brain involve unsupervised learning, and have been strongly tied to specific plasticity rules. To close this gap, we study an instantiation of simple-complex cell model and show, for a broad class of unsupervised learning rules (including Hebbian learning), that we can learn object representations that are invariant to nuisance transformations belonging to a finite orthogonal group. These findings may have implications for developing neurally plausible theories and models of how the visual cortex or artificial neural networks build selectivity for discriminating objects and invariance to real-world nuisance transformations.

A computational model for grid maps in neural populations.

Grid cells in the entorhinal cortex, together with head direction, place, speed and border cells, are major contributors to the organization of spatial representations in the brain. In this work we introduce a novel theoretical and algorithmic framework able to explain the optimality of hexagonal grid-like response patterns. We show that this pattern is a result of minimal variance encoding of neurons together with maximal robustness to neurons' noise and minimal number of encoding neurons. The novelty lies in the formulation of the encoding problem considering neurons as an overcomplete basis (a frame) where the position information is encoded. Through the modern Frame Theory language, specifically that of tight and equiangular frames, we provide new insights about the optimality of hexagonal grid receptive fields. The proposed model is based on the well-accepted and tested hypothesis of Hebbian learning, providing a simplified cortical-based framework that does not require the presence of velocity-driven oscillations (oscillatory model) or translational symmetries in the synaptic connections (attractor model). We moreover demonstrate that the proposed encoding mechanism naturally explains axis alignment of neighbor grid cells and maps shifts, rotations and scaling of the stimuli onto the shape of grid cells' receptive fields, giving a straightforward explanation of the experimental evidence of grid cells remapping under transformations of environmental cues.

View-Tolerant Face Recognition and Hebbian Learning Imply Mirror-Symmetric Neural Tuning to Head Orientation.

The primate brain contains a hierarchy of visual areas, dubbed the ventral stream, which rapidly computes object representations that are both specific for object identity and robust against identity-preserving transformations, like depth rotations [1, 2]. Current computational models of object recognition, including recent deep-learning networks, generate these properties through a hierarchy of alternating selectivity-increasing filtering and tolerance-increasing pooling operations, similar to simple-complex cells operations [3-6]. Here, we prove that a class of hierarchical architectures and a broad set of biologically plausible learning rules generate approximate invariance to identity-preserving transformations at the top level of the processing hierarchy. However, all past models tested failed to reproduce the most salient property of an intermediate representation of a three-level face-processing hierarchy in the brain: mirror-symmetric tuning to head orientation [7]. Here, we demonstrate that one specific biologically plausible Hebb-type learning rule generates mirror-symmetric tuning to bilaterally symmetric stimuli, like faces, at intermediate levels of the architecture and show why it does so. Thus, the tuning properties of individual cells inside the visual stream appear to result from group properties of the stimuli they encode and to reflect the learning rules that sculpted the information-processing system within which they reside.

The Invariance Hypothesis Implies Domain-Specific Regions in Visual Cortex.

Is visual cortex made up of general-purpose information processing machinery, or does it consist of a collection of specialized modules? If prior knowledge, acquired from learning a set of objects is only transferable to new objects that share properties with the old, then the recognition system's optimal organization must be one containing specialized modules for different object classes. Our analysis starts from a premise we call the invariance hypothesis: that the computational goal of the ventral stream is to compute an invariant-to-transformations and discriminative signature for recognition. The key condition enabling approximate transfer of invariance without sacrificing discriminability turns out to be that the learned and novel objects transform similarly. This implies that the optimal recognition system must contain subsystems trained only with data from similarly-transforming objects and suggests a novel interpretation of domain-specific regions like the fusiform face area (FFA). Furthermore, we can define an index of transformation-compatibility, computable from videos, that can be combined with information about the statistics of natural vision to yield predictions for which object categories ought to have domain-specific regions in agreement with the available data. The result is a unifying account linking the large literature on view-based recognition with the wealth of experimental evidence concerning domain-specific regions.

The diagnostic accuracy of five tests for diagnosing partial-thickness tears of the supraspinatus tendon: A cohort study.

STUDY DESIGN: A cohort study. INTRODUCTION: The causes of the pain can be difficult for clinicians to diagnose due to the complexity of the shoulder anatomy and the wide spectrum of shoulder conditions. PURPOSE OF THE STUDY: The aim of this study was to investigate the clinical usefulness of provocative diagnostic tests, in patients with partial-thickness tears of the supraspinatus (SST) tendon. METHODS: The partial-thickness tears SST tendon group consisted of 50 patients and 50 subjects with shoulder pain. Sensitivity, specificity, positive and negative predictive values, and likelihood ratios of five physical examination tests were calculated using a 2 × 2 table. RESULTS: All the tests exhibited clinical useful positive LR greater than 2 (all, >7.0). The Yocum, Jobe, and Hawkinse-Kennedy exhibited clinically useful negative LR of less than 0.5. CONCLUSIONS: The provocative tests examined were clinically useful in determining the presence or absence of pathology of the supraspinatus tendon. LEVEL OF EVIDENCE: 2b.

Adenosine triphosphate acts as a paracrine signaling molecule to reduce the motility of T cells.

Organization of immune responses requires exchange of information between cells. This is achieved through either direct cell-cell contacts and establishment of temporary synapses or the release of soluble factors, such as cytokines and chemokines. Here we show a novel form of cell-to-cell communication based on adenosine triphosphate (ATP). ATP released by stimulated T cells induces P2X4/P2X7-mediated calcium waves in the neighboring lymphocytes. Our data obtained in lymph node slices suggest that, during T-cell priming, ATP acts as a paracrine messenger to reduce the motility of lymphocytes and that this may be relevant to allow optimal tissue scanning by T cells.

The CXCR4 mutations in WHIM syndrome impair the stability of the T-cell immunologic synapse.

WHIM (warts, hypogammaglobulinemia, infections, myelokathexis) syndrome is a rare disease characterized by diverse symptoms indicative of aberrantly functioning immunity. It is caused by mutations in the chemokine receptor CXCR4, which impair its intracellular trafficking, leading to increased responsiveness to chemokine ligand and retention of neutrophils in bone marrow. Yet WHIM symptoms related to adaptive immunity, such as delayed IgG switching and impaired memory B-cell function, remain largely unexplained. We hypothesized that the WHIM-associated mutations in CXCR4 may affect the formation of immunologic synapses between T cells and antigen-presenting cells (APCs). We show that, in the presence of competing external chemokine signals, the stability of T-APC conjugates from patients with WHIM-mutant CXCR4 is disrupted as a result of impaired recruitment of the mutant receptor to the immunologic synapse. Using retrogenic mice that develop WHIM-mutant T cells, we show that WHIM-mutant CXCR4 inhibits the formation of long-lasting T-APC interactions in ex vivo lymph node slice time-lapse microscopy. These findings demonstrate that chemokine receptors can affect T-APC synapse stability and allow us to propose a novel mechanism that contributes to the adaptive immune response defects in WHIM patients.

Reduced phosphatidylinositol 4,5-bisphosphate synthesis impairs inner ear Ca2+ signaling and high-frequency hearing acquisition.

Phosphatidylinositol phosphate kinase type 1γ (PIPKIγ) is a key enzyme in the generation of phosphatidylinositol 4,5-bisphosphate [PI(4,5)P(2)] and is expressed at high levels in the nervous system. Homozygous knockout mice lacking this enzyme die postnatally within 24 h, whereas PIPKIγ(+/-) siblings breed normally and have no reported phenotype. Here we show that adult PIPKIγ(+/-) mice have dramatically elevated hearing thresholds for high-frequency sounds. During the first postnatal week we observed a reduction of ATP-dependent Ca(2+) signaling activity in cochlear nonsensory cells. Because Ca(2+) signaling under these conditions depends on inositol-1,4,5-trisphosphate generation from phospholipase C (PLC)-dependent hydrolysis of PI(4,5)P(2), we conclude that (i) PIPKIγ is primarily responsible for the synthesis of the receptor-regulated PLC-sensitive PI(4,5)P(2) pool in the cell syncytia that supports auditory hair cells; (ii) spatially graded impairment of this signaling pathway in cochlear nonsensory cells causes a selective alteration in the acquisition of hearing in PIPKIγ(+/-) mice. This mouse model also suggests that PIPKIγ may determine the level of gap junction contribution to cochlear development.

Adhesion shapes T cells for prompt and sustained T-cell receptor signalling.

During T-cell migration, cell polarity is orchestrated by chemokine receptors and adhesion molecules and involves the functional redistribution of molecules and organelles towards specific cell compartments. In contrast, it is generally believed that the cell polarity established when T cells meet antigen-presenting cells (APCs) is controlled by the triggered T-cell receptor (TCR). Here, we show that, during activation of human T lymphocytes by APCs, chemokines and LFA-1 establish cell polarity independently of TCR triggering. Chemokine-induced LFA-1 activation results in fast recruitment of MTOC and mitochondria towards the potential APC, a process required to amplify TCR Ca(2+) signalling at the upcoming immunological synapse, to promote nuclear translocation of transcriptional factor NFATc2 and boost CD25 expression. Our data show that the initial adhesive signals delivered by chemokines and LFA-1 shape and prepare T cells for antigen recognition.

ATP-mediated cell-cell signaling in the organ of Corti: the role of connexin channels.

UNLABELLED: Connexin 26 (Cx26) and connexin 30 (Cx30) form hemichannels that release ATP from the endolymphatic surface of cochlear supporting and epithelial cells and also form gap junction (GJ) channels that allow the concomitant intercellular diffusion of Ca(2+) mobilizing second messengers. Released ATP in turn activates G-protein coupled P2Y(2) and P2Y(4) receptors, PLC-dependent generation of IP(3), release of Ca(2+) from intracellular stores, instigating the regenerative propagation of intercellular Ca(2+) signals (ICS). The range of ICS propagation is sensitive to the concentration of extracellular divalent cations and activity of ectonucleotidases. Here, the expression patterns of Cx26 and Cx30 were characterized in postnatal cochlear tissues obtained from mice aged between P5 and P6. The expression gradient along the longitudinal axis of the cochlea, decreasing from the basal to the apical cochlear turn (CT), was more pronounced in outer sulcus (OS) cells than in inner sulcus (IS) cells. GJ-mediated dye coupling was maximal in OS cells of the basal CT, inhibited by the nonselective connexin channel blocker carbenoxolone (CBX) and absent in hair cells. Photostimulating OS cells with caged inositol (3,4,5) tri-phosphate (IP(3)) resulted in transfer of ICS in the lateral direction, from OS cells to IS cells across the hair cell region (HCR) of medial and basal CTs. ICS transfer in the opposite (medial) direction, from IS cells photostimulated with caged IP(3) to OS cells, occurred mostly in the basal CT. In addition, OS cells displayed impressive rhythmic activity with oscillations of cytosolic free Ca(2+) concentration ([Ca(2+)](i)) coordinated by the propagation of Ca(2+) wavefronts sweeping repeatedly through the same tissue area along the coiling axis of the cochlea. Oscillations evoked by uncaging IP(3) or by applying ATP differed greatly, by as much as one order of magnitude, in frequency and waveform rise time. ICS evoked by direct application of ATP propagated along convoluted cellular paths in the OS, which often branched and changed dynamically over time. Potential implications of these findings are discussed in the context of developmental regulation and cochlear pathophysiology. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (doi:10.1007/s11302-010-9192-9) contains supplementary material, which is available to authorized users.

ATP release through connexin hemichannels and gap junction transfer of second messengers propagate Ca2+ signals across the inner ear.

Extracellular ATP controls various signaling systems including propagation of intercellular Ca(2+) signals (ICS). Connexin hemichannels, P2x7 receptors (P2x7Rs), pannexin channels, anion channels, vesicles, and transporters are putative conduits for ATP release, but their involvement in ICS remains controversial. We investigated ICS in cochlear organotypic cultures, in which ATP acts as an IP(3)-generating agonist and evokes Ca(2+) responses that have been linked to noise-induced hearing loss and development of hair cell-afferent synapses. Focal delivery of ATP or photostimulation with caged IP(3) elicited Ca(2+) responses that spread radially to several orders of unstimulated cells. Furthermore, we recorded robust Ca(2+) signals from an ATP biosensor apposed to supporting cells outside the photostimulated area in WT cultures. ICS propagated normally in cultures lacking either P2x7R or pannexin-1 (Px1), as well as in WT cultures exposed to blockers of anion channels. By contrast, Ca(2+) responses failed to propagate in cultures with defective expression of connexin 26 (Cx26) or Cx30. A companion paper demonstrates that, if expression of either Cx26 or Cx30 is blocked, expression of the other is markedly down-regulated in the outer sulcus. Lanthanum, a connexin hemichannel blocker that does not affect gap junction (GJ) channels when applied extracellularly, limited the propagation of Ca(2+) responses to cells adjacent to the photostimulated area. Our results demonstrate that these connexins play a dual crucial role in inner ear Ca(2+) signaling: as hemichannels, they promote ATP release, sustaining long-range ICS propagation; as GJ channels, they allow diffusion of Ca(2+)-mobilizing second messengers across coupled cells.

Coordinated control of connexin 26 and connexin 30 at the regulatory and functional level in the inner ear.

Connexin 26 (Cx26) and connexin 30 (Cx30) are encoded by two genes (GJB2 and GJB6, respectively) that are found within 50 kb in the same complex deafness locus, DFNB1. Immunocytochemistry and quantitative PCR analysis of Cx30 KO mouse cultures revealed that Cx26 is downregulated at the protein level and at the mRNA level in nonsensory cells located between outer hair cells and the stria vascularis. To explore connexin coregulation, we manipulated gene expression using the bovine adeno-associated virus. Overexpression of Cx30 in the Cx30 KO mouse by transduction with bovine adeno-associated virus restored Cx26 expression, permitted the formation of functional gap junction channels, and rescued propagating Ca(2+) signals. Ablation of Cx26 by transduction of Cx26(loxP/loxP) cultures with a Cre recombinase vector caused concurrent downregulation of Cx30 and impaired intercellular communication. The coordinated regulation of Cx26 and Cx30 expression appears to occur as a result of signaling through PLC and the NF-kappaB pathway, because activation of IP(3)-mediated Ca(2+) responses by stimulation of P2Y receptors for 20 min with 20 nM ATP increased the levels of Cx26 transcripts in Cx30 KO cultures. This effect was inhibited by expressing a stable form of the IkappaB repressor protein that prevents activation/translocation of NF-kappaB. Thus, our data reveal a Ca(2+)-dependent control in the expression of inner ear connexins implicated in hereditary deafness as well as insight into the hitherto unexplained observation that some deafness-associated DFNB1 alleles are characterized by hereditable reduction of both GJB2 and GJB6 expression.

Ca2+ signaling in the inner ear.

The inner ear contains delicate sensory receptors that have adapted to detect the minutest mechanical disturbances. Ca(2+) ions are implicated in all steps of the transduction process, as well as in its regulation by an impressive ensemble of finely tuned feedback control mechanisms. Recent studies have unveiled some of the key players, but things do not sound quite right yet.