Background Glioblastoma (GBM), the most aggressive primary brain tumor, has a median survival of approximately 15 months. Twenty percent of patients survive beyond three years, but known clinical factors like age, performance status, resection extent, and MGMT promoter methylation status do not fully explain the observed outcomes. Objective Our objective was to identify novel histology derived biomarkers associated with end-of-spectrum overall survival (OS) to provide novel biological insight with a translational potential. Methods We analyzed a total of 748 GBM patients from 3 different cohorts, uniquely enriched in long survivors (n=98 with overall survival (OS) > 5y including n=196 with OS≥3y), with clinical data and H&E slides obtained from the primary tumor at baseline. We propose an interpretable machine learning (ML) methodology for the discovery of histological biomarkers. Our method learned to segment each H&E slide into three distinct regions associated with long-term survival, short-term survival, and non-informative tissue. We characterized these regions by integrating unsupervised learning, nuclei segmentation, blood vessels detection, pathologist annotations, and multimodal data including spatial transcriptomics from n=31 patients of the GBM MOSAIC dataset to discover fully interpretable biomarkers. Results Our OS prediction model using histology and clinical data as input achieved an area under the curve (AUC) of 0.85 for the classification of patients between OS<2 and OS≥3y in external cohort validation, outperforming significantly models trained on clinical data or on histology alone (AUC of 0.76; 0.73, respectively). Two novel biomarkers were predicting poor survival: the presence of regions of lowly infiltrated white matter enriched in malignant cells with a mesenchymal-like phenotype, and lower levels of angiogenesis associated with higher hypoxia response in the main tumor regions. We also found that a subtype of immunosuppressive tumor macrophages - defined by high PLIN2 expression and lipid accumulation- is consistently enriched in histological areas predictive of poor prognosis. Conclusion Our interpretable ML methodology identified a novel prognostic impact of biological processes and cell types according to distinct tumor regions of GBM. These results pave the way for spatially-informed biomarkers to improve risk stratification and for personalized spatially-targeted therapeutic strategies. Key highlights 1. Our ML model identified histological biomarkers predicting prognosis independently from known clinical factors 2. The region of lowly infiltrated white matter enriched in malignant cells including a mesenchymal-like phenotype is predictive of poor prognosis 3. Angiogenesis is increased in areas predictive of long survival in main non-necrotic tumor regions. 4. The subtype of macrophages expressing PLIN2 and associated with increased lipid metabolism was associated with poor prognosis in all GBM regions. ![Figure][1] ### Competing Interest Statement The authors have declared no competing interest. [1]: pending:yes

Cell-cell communication is central to understanding the complex interactions within the tumor microenvironment. However, current methods fail to identify recurrent communication patterns across patient cohorts from spatial transcriptomics, as they are often limited to single samples or lack essential spatial context. Yet this is essential for understanding how local environments influence cell phenotype and states, and shape the entire cellular ecosystem. We introduce a machine-learning approach that models local, spatially aware ligand-receptor interactions and uses matrix factorization to extract global multicellular programs from large cohorts representing the complex biology of cancer. Applied to a multimodal muscle-invasive bladder cancer cohort of 146 patients, it uncovered 45 communication programs defined by distinct ligand-receptor pairs and cellular niches. In particular, we identified a conserved immune program linked to stalled anti-tumor immunity and a program linking KMT2D loss-of-function mutations with early-stage (T2) tumors, intense proliferation and a favorable response to neoadjuvant chemotherapy. ### Competing Interest Statement The authors have declared no competing interest.

Abstract Hippo signaling emerged over the last decade as a major tumor-suppressing pathway. Its dysregulation is generally associated with abnormal expression levels of YAP1, WWTR1 (coding for the TAZ protein) and TEAD genes among others. This pathway has been shown to have a prognostic impact in several cancer types. In particular, the role of YAP1/TEAD activity across indications has been emphasized by several recent works, with potential implications on treatment options. Therefore, identifying patients with a deregulated pathway is key for a better clinical impact of the current potential therapies. Recent studies have been able to characterize RNA-seq based signatures of a deregulated Hippo pathway but a reproducible and cost-effective method to measure the activation of the pathway in clinical settings is needed. Here we first evaluate and confirm the robustness of a YAP1/TEAD activity signature recently published by Calvet et al. (RNA-based signature focused on TEAD downstream effectors), to predict the level of activity of this pathway across several cancer types available in TCGA. Our results confirmed that YAP1/TEAD activity is cancer type and subtype specific and that its high activity is correlated with poor prognosis in some of these cancers. We then trained deep learning models to predict YAP1/TEAD gene activation level, from H&E-stained histology slides in various cancer types from The Cancer Genome Atlas. We showed that histological markers associated with dysfunctional Hippo signaling are markers of disease aggressiveness and poor prognosis such as necrosis, poorly differentiated tumor, and inflammation. Altogether our results are opening the avenue of defining image-based biomarkers predictive of the YAP1/TEAD activity which could be used in clinical settings for better inclusion of subgroups of patients for targeted therapeutics against this hallmark of cancer development. Citation Format: Benoit Schmauch, Vincent Cabeli, Omar Darwiche-Domingues, Jean-Eudes Le Douget, Alexandra Hardy, Reda Belbahri, Charles Maussion, Alberto Romagnoni, Markus Eckstein, Florian Fuchs, Aurélie Swalduz, Sylvie Lantuejoul, Hugo Crochet, François Ghiringhelli, Valentin Derangere, Caroline Truntzer, Harvey Pass, Andre Moreira, Luis Chiriboga, Yuanning Zheng, Michael Ozawa, Brooke Howitt, Olivier Gevaert, Nicolas Girard, Elton Rexhepaj, Iris Valtingojer, Laurent Debussche, Eric Durand, Marion Classe, Katharina Von Loga, Elodie Pronier, Matteo Cesaroni. Deep learning uncovers morphological patterns of YAP1/TEAD activity related to disease aggressiveness in cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7377.

Deep learning (DL) has shown potential to provide powerful representations of bulk RNA-seq data in cancer research. However, there is no consensus regarding the impact of design choices of DL approaches on the performance of the learned representation, including the model architecture, the training methodology and the various hyperparameters. To address this problem, we evaluate the performance of various design choices of DL representation learning methods using TCGA and DepMap pan-cancer datasets and assess their predictive power for survival and gene essentiality predictions. We demonstrate that baseline methods achieve comparable or superior performance compared to more complex models on survival predictions tasks. DL representation methods, however, are the most efficient to predict the gene essentiality of cell lines. We show that auto-encoders (AE) are consistently improved by techniques such as masking and multi-head training. Our results suggest that the impact of DL representations and of pretraining are highly task- and architecture-dependent, highlighting the need for adopting rigorous evaluation guidelines. These guidelines for robust evaluation are implemented in a pipeline made available to the research community.

Over the last decade, Hippo signaling has emerged as a major tumor-suppressing pathway. Its dysregulation is associated with abnormal expression of YAP1 and TEAD-family genes. Recent works have highlighted the role of YAP1/TEAD activity in several cancers and its potential therapeutic implications. Therefore, identifying patients with a dysregulated Hippo pathway is key to enhancing treatment impact. Although recent studies have derived RNA-seq-based signatures, there remains a need for a reproducible and cost-effective method to measure the pathway activation. In recent years, deep learning applied to histology slides have emerged as an effective way to predict molecular information from a data modality available in clinical routine. Here, we trained models to predict YAP1/TEAD activity from H&E-stained histology slides in multiple cancers. The robustness of our approach was assessed in seven independent validation cohorts. Finally, we showed that histological markers of disease aggressiveness were associated with dysfunctional Hippo signaling.

Neuronal cable theory is usually derived from an electric analogue of the membrane, which contrasts with the slow movement of ions in aqueous media. We show here that it is possible to derive neuronal cable equations from a different perspective, based on the laws of hydrodynamic motion of charged particles (Navier-Stokes equations). This results in similar cable equations, but with additional contributions arising from nonlinear interactions inherent to fluid dynamics, and which may shape the integrative properties of the neurons.

The forward BFKL equation is discretised in virtuality space and it is shown that the diffusion into infrared and ultraviolet momenta can be understood in terms of a semi-infinite matrix. The square truncation of this matrix can be exponentiated leading to asymptotic eigenstates sharing many features with the BFKL gluon Green’s function in the limit of large matrix size. This truncation is closely related to a representation of the XXX Heisenberg spin = - 1/2 chain with SL(2) invariance where the Hamiltonian acts on a symmetric double copy of the harmonic oscillator. A simple modification of the BFKL matrix suppressing the infrared modes generates evolution more compatible with the Froissart bound.

Nervous system maturation occurs on multiple levels, synaptic, circuit, and network, at divergent time scales. For example, many synaptic properties mature gradually, while emergent network dynamics, as data show, change abruptly. Here, we combine experimental and theoretical approaches to investigate a sudden transition in spontaneous thalamocortical activity necessary for the development of vision. Inspired by in vivo measurements of time-scales and amplitudes of synaptic currents, we extend the Wilson and Cowan model to take into account the relative onset timing and amplitudes of inhibitory and excitatory neural population responses. We study the dynamics of this system and identify the bifurcations as the onset timescales of excitation and inhibition are varied. We focus on the specific typical developmental changes in synaptic timescales consistent with the experimental observations. These findings argue that the inhibitory timing is a critical determinant of thalamocortical activity maturation; a gradual decay of the ratio of inhibitory to excitatory onset time below one drives the system through a bifurcation that leads to a sudden switch of the network spontaneous activity from high-amplitude oscillations to a non-oscillatory active state. This switch also drives a marked change to a linear network response to transient stimuli, agreeing to the in vivo observations. The switch observed in the model is representative of the sudden transition in the sensory cortical activity seen early in development.

Deep learning methods for digital pathology analysis have proved an effective way to address multiple clinical questions, from diagnosis to prognosis and even to prediction of treatment outcomes. They have also recently been used to predict gene mutations from pathology images, but no comprehensive evaluation of their potential for extracting molecular features from histology slides has yet been performed. We propose a novel approach based on the integration of multiple data modes, and show that our deep learning model, HE2RNA, can be trained to systematically predict RNA-Seq profiles from whole-slide images alone, without the need for expert annotation. HE2RNA is interpretable by design, opening up new opportunities for virtual staining. In fact, it provides virtual spatialization of gene expression, as validated by double-staining on an independent dataset. Moreover, the transcriptomic representation learned by HE2RNA can be transferred to improve predictive performance for other tasks, particularly for small datasets. As an example of a task with direct clinical impact, we studied the prediction of microsatellite instability from hematoxylin & eosin stained images and our results show that better performance can be achieved in this setting.

Deep learning methods for digital pathology analysis are an effective way to address multiple clinical questions, from diagnosis to prediction of treatment outcomes. These methods have also been used to predict gene mutations from pathology images, but no comprehensive evaluation of their potential for extracting molecular features from histology slides has yet been performed. We show that HE2RNA, a model based on the integration of multiple data modes, can be trained to systematically predict RNA-Seq profiles from whole-slide images alone, without expert annotation. Through its interpretable design, HE2RNA provides virtual spatialization of gene expression, as validated by CD3- and CD20-staining on an independent dataset. The transcriptomic representation learned by HE2RNA can also be transferred on other datasets, even of small size, to increase prediction performance for specific molecular phenotypes. We illustrate the use of this approach in clinical diagnosis purposes such as the identification of tumors with microsatellite instability.

More interest has been shown in recent years to large-scale spiking simulations of cerebral neuronal networks, coming both from the presence of high-performance computers and increasing details in experimental observations. In this context it is important to understand how population dynamics are generated by the designed parameters of the networks, which is the question addressed by mean-field theories. Despite analytic solutions for the mean-field dynamics already being proposed for current-based neurons (CUBA), a complete analytic description has not been achieved yet for more realistic neural properties, such as conductance-based (COBA) network of adaptive exponential neurons (AdEx). Here, we propose a principled approach to map a COBA on a CUBA. Such an approach provides a state-dependent approximation capable of reliably predicting the firing-rate properties of an AdEx neuron with noninstantaneous COBA integration. We also applied our theory to population dynamics, predicting the dynamical properties of the network in very different regimes, such as asynchronous irregular and synchronous irregular (slow oscillations). This result shows that a state-dependent approximation can be successfully introduced to take into account the subtle effects of COBA integration and to deal with a theory capable of correctly predicting the activity in regimes of alternating states like slow oscillations.

Accurate population models are needed to build very large-scale neural models, but their derivation is difficult for realistic networks of neurons, in particular when nonlinear properties are involved, such as conductance-based interactions and spike-frequency adaptation. Here, we consider such models based on networks of adaptive exponential integrate-and-fire excitatory and inhibitory neurons. Using a master equation formalism, we derive a mean-field model of such networks and compare it to the full network dynamics. The mean-field model is capable of correctly predicting the average spontaneous activity levels in asynchronous irregular regimes similar to in vivo activity. It also captures the transient temporal response of the network to complex external inputs. Finally, the mean-field model is also able to quantitatively describe regimes where high- and low-activity states alternate (up-down state dynamics), leading to slow oscillations. We conclude that such mean-field models are biologically realistic in the sense that they can capture both spontaneous and evoked activity, and they naturally appear as candidates to build very large-scale models involving multiple brain areas.

Crohn Disease (CD) is a complex genetic disorder for which more than 140 genes have been identified using genome wide association studies (GWAS). However, the genetic architecture of the trait remains largely unknown. The recent development of machine learning (ML) approaches incited us to apply them to classify healthy and diseased people according to their genomic information. The Immunochip dataset containing 18,227 CD patients and 34,050 healthy controls enrolled and genotyped by the international Inflammatory Bowel Disease genetic consortium (IIBDGC) has been re-analyzed using a set of ML methods: penalized logistic regression (LR), gradient boosted trees (GBT) and artificial neural networks (NN). The main score used to compare the methods was the Area Under the ROC Curve (AUC) statistics. The impact of quality control (QC), imputing and coding methods on LR results showed that QC methods and imputation of missing genotypes may artificially increase the scores. At the opposite, neither the patient/control ratio nor marker preselection or coding strategies significantly affected the results. LR methods, including Lasso, Ridge and ElasticNet provided similar results with a maximum AUC of 0.80. GBT methods like XGBoost, LightGBM and CatBoost, together with dense NN with one or more hidden layers, provided similar AUC values, suggesting limited epistatic effects in the genetic architecture of the trait. ML methods detected near all the genetic variants previously identified by GWAS among the best predictors plus additional predictors with lower effects. The robustness and complementarity of the different methods are also studied. Compared to LR, non-linear models such as GBT or NN may provide robust complementary approaches to identify and classify genetic markers.

Emotional disorders and psychological flourishing are the result of complex interactions between positive and negative affects that depend on external events and the subject's internal representations. Based on psychological data, we mathematically model the dynamical balance between positive and negative affects as a function of the response to external positive and negative events. This modeling allows the investigation of the relative impact of two leading forms of therapy on affect balance. The model uses a delay differential equation to analytically study the bifurcation diagram of the system. We compare the results of the model to psychological data on a single, recurrently depressed patient who was administered the two types of therapies considered (coping focused versus affect focused). The model leads to the prediction that stabilization at a normal state may rely on evaluating one's emotional state through a historical ongoing emotional state rather than in a narrow present window. The simple mathematical model proposed here offers a theoretical framework for investigating the temporal process of change and parameters of resilience to relapse.

In the early visual cortex, information is processed within functional maps whose layouts are thought to underlie visual perception. However, the precise organization of these functional maps as well as their interrelationships remain unsettled. Here, we show that spatial frequency representation in cat early visual cortex exhibits singularities around which the map organizes like an electric dipole potential. These singularities are precisely co-located with singularities of the orientation map: the pinwheel centers. To show this, we used high resolution intrinsic optical imaging in cat areas 17 and 18. First, we show that a majority of pinwheel centers exhibit in their neighborhood both semi-global maximum and minimum in the spatial frequency map (i.e. extreme values of the spatial frequency in a hypercolumn). This contradicts pioneering studies suggesting that pinwheel centers are placed at the locus of a single spatial frequency extremum. Based on an analogy with electromagnetism, we proposed a mathematical model for a dipolar structure, accurately fitting optical imaging data. We conclude that a majority of orientation pinwheel centers form spatial frequency dipoles in cat early visual cortex. Given the functional specificities of neurons at singularities in the visual cortex, it is argued that the dipolar organization of spatial frequency around pinwheel centers could be fundamental for visual processing.

The layout of sensory brain areas is thought to subtend perception. The principles shaping these architectures and their role in information processing are still poorly understood. We investigate mathematically and computationally the representation of orientation and spatial frequency in cat primary visual cortex. We prove that two natural principles, local exhaustivity and parsimony of representation, would constrain the orientation and spatial frequency maps to display a very specific pinwheel-dipole singularity. This is particularly interesting since recent experimental evidences show a dipolar structures of the spatial frequency map co-localized with pinwheels in cat. These structures have important properties on information processing capabilities. In particular, we show using a computational model of visual information processing that this architecture allows a trade-off in the local detection of orientation and spatial frequency, but this property occurs for spatial frequency selectivity sharper than reported in the literature. We validated this sharpening on high-resolution optical imaging experimental data. These results shed new light on the principles at play in the emergence of functional architecture of cortical maps, as well as their potential role in processing information.

In the minimal supersymmetric standard model (MSSM), the Higgs couplings to down-type quarks and leptons, normalized with respect to their corresponding Standard Model values, coincide at tree level and this degeneracy is only slightly broken at the quantum level. Motivated by the latest results of the Higgs searches at the LHC and Tevatron, we explore the possibility of disentangling these couplings from each other by considering a scenario in which supersymmetry is broken spontaneously at a low scale. In such a scenario, all the Higgs couplings, except the ones to the Z and W bosons, receive tree level corrections that depend on the MSSM soft parameters. In particular, the corrections to the Yukawa couplings depend on the A-term soft parameter for the corresponding fermion, allowing for the freedom to break their usual relations, even in the MSSM decoupling limit. We highlight the main features of this scenario in terms of a benchmark point for which the normalized Higgs coupling to the tau leptons is depleted, the coupling to photons is enhanced, while all the other Higgs couplings, including the one to the bottom quarks, are close to their corresponding standard model values. We also discuss the experimental bounds arising mainly from di-tau searches and comment on the discovery/exclusion prospects at the LHC. DOI: 10.1103/PhysRevD.87.013008

Extensions of the Standard Model with an extra U'(1) abelian group generically generate terms coming from loops of heavy fermions, leading to three gauge boson couplings, in particular Z'Z gamma. We show that WMAP data constrains the gauge coupling of the group g_D to values comparable with the electro-weak ones, rather independently of the mass of Z'. Moreover, the model predicts a monochromatic gamma-ray line which can fit a 130 GeV signal at the FERMI telescope for natural values of the Chern-Simons terms and a dark matter mass around 144.5 GeV.