Biology subjects

Levy, J.

Publications and source records attributed to Levy, J..

36 records · Page 2Linked to original sources

bioRxiv2022

Uncovering Additional Predictors of Urothelial Carcinoma from Voided Urothelial Cell Clusters Through a Deep Learning Based Image Preprocessing Technique

Urine cytology is commonly used as a screening test for high grade urothelial carcinoma for patients with risk factors or hematuria and is an essential step in longitudinal monitoring of patients with a prior bladder cancer history. However, the semi-subjective nature of current reporting systems for urine cytology (e.g., The Paris System) can hamper reproducibility. For instance, the incorporation of urothelial cell clusters into the classification schema is still an item of debate and perplexity amongst expert cytopathologists, as several previous works have disputed their diagnostic relevance. Recently, several machine learning and morphometric algorithms have been proposed to provide quantitative descriptors of urine cytology specimens in an effort to reduce subjectivity and include automated assessments of cell clusters. However, it remains unclear how these computer algorithms interpret/analyze cell clusters. In this work, we have developed an automated preprocessing tool for urothelial cell cluster assessment that divides urothelial cell clusters into meaningful components for downstream assessment (i.e., population-based studies, workflow automation). Results indicate that cell cluster atypia (i.e., defined by whether the cell cluster harbored multiple atypical cells, thresholded by a minimum number of cells), cell border overlap and smoothness, and total number of clusters are important markers of specimen atypia when considering assessment of urothelial cell clusters. Markers established through techniques to separate cell clusters may have wider applicability for the design and implementation of machine learning approaches for urine cytology assessment.

pathology↗

bioRxiv2021

Video-Based Deep Learning to Detect Dyssynergic Defecation with 3D High-Definition Anorectal Manometry

BACKGROUND AND AIMSEvaluation for dyssynergia is the most common reason that gastroenterologists refer patients for anorectal manometry, because dyssynergia is amenable to biofeedback by physical therapists. High-definition anorectal manometry (3D-HDAM) is a promising technology to evaluate anorectal physiology, but adoption remains limited by its sheer complexity. We developed a 3D-HDAM deep learning algorithm to evaluate for dyssynergia. METHODSSpatial-temporal data were extracted from consecutive 3D-HDAM studies performed between 2018-2020 at a tertiary institution. The technical procedure and gold standard definition of dyssynergia were based on the London consensus, adapted to the needs of 3D-HDAM technology. Three machine learning models were generated: (1) traditional machine learning informed by conventional anorectal function metrics, (2) deep learning, and (3) a hybrid approach. Diagnostic accuracy was evaluated using bootstrap sampling to calculate area-under-the-curve (AUC). To evaluate overfitting, models were validated by adding 502 simulated defecation maneuvers with diagnostic ambiguity. RESULTS302 3D-HDAM studies representing 1,208 simulated defecation maneuvers were included (average age 55.2 years; 80.5% women). The deep learning model had comparable diagnostic accuracy (AUC=0.91 [95% confidence interval 0.89-0.93]) to traditional (AUC=0.93[0.92-0.95]) and hybrid (AUC=0.96[0.94-0.97]) predictive models in training cohorts. However, the deep learning model handled ambiguous tests more cautiously than other models; the deep learning model was more likely to designate an ambiguous test as inconclusive (odds ratio=4.21[2.78-6.38]) versus traditional/hybrid approaches. CONCLUSIONSBy considering complex spatial-temporal information beyond conventional anorectal function metrics, deep learning on 3D-HDAM technology may enable gastroenterologists to reliably identify and manage dyssynergia in broader practice.

bioinformatics↗

bioRxiv2021

Cerebellar and subcortical atrophy contribute to psychiatric symptoms in frontotemporal dementia

Recent studies have suggested that cerebellar and subcortical structures are impacted early in the disease progression of genetic frontotemporal dementia (FTD) due to microtubule-associated protein tau (MAPT), progranulin (GRN) and chromosome 9 open reading frame 72 (C9orf72). However, the clinical contribution of the structures involved in the cerebello-subcortical circuitry has been understudied in FTD given their potentially central role in cognition and behaviour processes. The present study aims to investigate whether there is an association between the atrophy of the cerebellar and subcortical structures, and neuropsychiatric symptoms (using the revised version of the Cambridge Behavioral Inventory, CBI-R) across genetic mutations and whether this association starts during the preclinical phase of the disease. Our study included 983 participants from the Genetic Frontotemporal dementia Initiative (GENFI) including mutation carriers (n=608) and non-carrier first-degree relatives of known symptomatic carriers (n= 375). Voxel-wise analysis of the thalamus, striatum, globus pallidus, amygdala, and the cerebellum was performed using deformation based morphometry (DBM) and partial least squares analyses (PLS) were used to link morphometry and behavioural symptoms. Our univariate results suggest that in this group of primarily presymptomatic subjects, volume loss in subcortical and cerebellar structure was primarily a function of aging, with only the C9orf72 group showing more pronounced volume loss in the thalamus compared to the non-carrier individuals. PLS analyses demonstrated that the cerebello-subcortical circuitry is related to all neuropsychiatric symptoms from the CBI-R, with significant overlap in brain/behaviour patterns, but also specificity for each genetic group. The biggest differences were in the extent of the cerebellar involvement (larger extent in C9orf72 group) and more prominent amygdalar contribution in the MAPT group. Finally, our findings demonstrated that C9orf72 and MAPT brain scores were related to estimated years before the age of symptom onset (EYO) in a second order relationship highlighting a steeper brain score decline 20 years before expected symptom onset, while GRN brain scores were related to age and not EYO. Overall, these results demonstrated the important role of the subcortical structures and especially of the cerebellum in genetic FTD symptom expression.

neuroscience↗

bioRxiv2021

Development of Biologically Interpretable Multimodal Deep Learning Model for Cancer Prognosis Prediction

Robust cancer prognostication can enable more effective patient care and management, which may potentially improve health outcomes. Deep learning has proven to be a powerful tool to extract meaningful information from cancer patient data. In recent years it has displayed promise in quantifying prognostication by predicting patient risk. However, most current deep learning-based cancer prognosis prediction methods use only a single data source and miss out on learning from potentially rich relationships across modalities. Existing multimodal approaches are challenging to interpret in a biological or medical context, limiting real-world clinical integration as a trustworthy prognostic decision aid. Here, we developed a multimodal modeling approach that can integrate information from the central modalities of gene expression, DNA methylation, and histopathological imaging with clinical information for cancer prognosis prediction. Our multimodal modeling approach combines pathway and gene-based sparsely coded layers with patch-based graph convolutional networks to facilitate biological interpretation of the model results. We present a preliminary analysis that compares the potential applicability of combining all modalities to uni- or bi-modal approaches. Leveraging data from four cancer subtypes from the Cancer Genome Atlas, results demonstrate the encouraging performance of our multimodal approach (C-index=0.660 without clinical features; C-index=0.665 with clinical features) across four cancer subtypes versus unimodal approaches and existing state-of-the-art approaches. This work brings insight to the development of interpretable multimodal methods of applying AI to biomedical data and can potentially serve as a foundation for clinical implementations of such software. We plan to follow up this preliminary analysis with an in-depth exploration of factors to improve multimodal modeling approaches on an in-house dataset.

pathology↗

bioRxiv2021

Configuring a federated network of real-world patient health data for multimodal deep learning prediction of health outcomes.

Vast quantities of electronic patient medical data are currently being collated and processed in large federated data repositories. For instance, TriNetX, Inc., a global health research network, has access to more than 300 million patients, sourced from healthcare organizations, biopharmaceutical companies, and contract research organizations. As such, pipelines that are able to algorithmically extract huge quantities of patient data from multiple modalities present opportunities to leverage machine learning and deep learning approaches with the possibility of generating actionable insight. In this work, we present a modular, semi-automated end-to-end machine and deep learning pipeline designed to interface with a federated network of structured patient data. This proof-of-concept pipeline is disease-agnostic, scalable, and requires little domain expertise and manual feature engineering in order to quickly produce results for the case of a user-defined binary outcome event. We demonstrate the pipelines efficacy with three different disease workflows, with high discriminatory power achieved in all cases.

bioinformatics↗

bioRxiv2021

Programmable large DNA deletion, replacement, integration, and inversion with twin prime editing and site-specific recombinases

The targeted deletion, replacement, integration, or inversion of DNA sequences at specified locations in the genome could be used to study or treat many human genetic diseases. Here, we describe twin prime editing (twinPE), a method for the programmable replacement or excision of DNA sequence at endogenous human genomic sites without requiring double-strand DNA breaks. TwinPE uses a prime editor (PE) protein and two prime editing guide RNAs (pegRNAs) that template the synthesis of complementary DNA flaps on opposing strands of genomic DNA, resulting in the replacement of endogenous DNA sequence between the PE-induced nick sites with pegRNA-encoded sequences. We show that twinPE in human cells can perform precise deletions of at least 780 bp and precise replacements of genomic DNA sequence with new sequences of at least 108 bp. By combining single or multiplexed twinPE with site-specific serine recombinases either in separate steps or in a single step, we demonstrate targeted integration of gene-sized DNA plasmids (>5,000 bp) into safe-harbor loci including AAVS1, CCR5, and ALB in human cells. To our knowledge, these results represent the first RNA-programmable insertion of gene-sized DNA sequences into targeted genomic sites of unmodified human cells without requiring double-strand breaks or homology-directed repair. Twin PE combined with recombinases also mediated a 40,167-bp inversion at IDS that corrects a common Hunter syndrome allele. TwinPE expands the capabilities of precision gene editing without requiring double-strand DNA breaks and synergizes with other tools to enable the correction or complementation of large or complex pathogenic alleles in human cells.

genomics↗

bioRxiv2021

Improving the Virtual Trichrome Assessment through Bridge Category Models

Non-alcoholic steatohepatitis (NASH) is a liver disease characterized by excessive lipid accumulation and disease progression is typically assessed through inspection of a Trichrome stain for Fibrosis staging. As the public health burden of NASH worsens due to evolving lifestyle habits, pathology laboratory resources will become increasingly strained due to rising demand for specialized stains. Virtual staining processes, computational methods which can synthesize the application of chemical staining reagents, can potentially provide resource savings by obviating the need to acquire specialized stains. Virtual staining technologies are assessed by comparing virtual and real tissue stains for their realism and ability to stage. However, these assessment methods are rife with statistical mistreatment of observed phenomena that are difficult to account for. Bridge category ratings represent a phenomenon where a pathologist may assign two adjacent stages simultaneously, which may bias and/or reduce the power of research findings. Such stage assignments were frequently reported in a large-scale assessment of Virtual Trichrome technologies yet were unaccounted for since no statistical adjustment procedures existed. In this work, we provide an updated assessment of Virtual Trichrome technologies using Bridge Category Models, which account for these bridge ratings. We report that two of four pathologists tended to assign lower Fibrosis stages to virtually stained tissue while the other two pathologists assigned similar stages. These research findings differ when bridge ratings are not accounted for. While promising, these results indicate further room for algorithmic finetuning of Virtual Trichrome technologies.

pathology↗

bioRxiv2021

Estimating the Inter- and Intra-Rater Reliability for NASH Fibrosis Staging in the Presence of Bridge Ordinal Ratings with Hierarchical Bridge Category Models

The public health burden of non-alcoholic steatohepatitis (NASH), a liver condition characterized by excessive lipid accumulation and subsequent tissue inflammation and fibrosis, has burgeoned with the spread of western lifestyle habits. Progression of fibrosis into cirrhosis is assessed using histological staging scales (e.g., NASH Clinical Research Network (NASH CRN)). These scales are used to monitor disease progression as well as to evaluate the effectiveness of therapies. However, clinical drug trials for NASH are typically underpowered due to lower than expected inter-/intra-rater reliability, which impacts measurements at screening, baseline, and endpoint. Bridge ratings represent a phenomenon where pathologists assign two adjacent stages simultaneously during assessment and may further complicate these analyses when ad hoc procedures are applied. Statistical techniques, dubbed Bridge Category Models, have been developed to account for bridge ratings, but not for the scenario where multiple pathologists assess biopsies across time points. Here, we develop hierarchical Bayesian extensions for these statistical methods to account for repeat observations and use these methods to assess the impact of bridge ratings on the inter-/intra-rater reliability of the NASH CRN staging scale. We also report on how pathologists may differ in their assignment of bridge ratings to highlight different staging practices. Our findings suggest that Bridge Category Models can capture additional fibrosis staging heterogeneity with greater precision, which translates to potentially higher reliability estimates in contrast to the information lost through ad hoc approaches.

pathology↗

bioRxiv2021

Bridge Category Models: Development of Bayesian Modelling Procedures to Account for Bridge Ordinal Ratings for Disease Staging

Disease grading and staging is accomplished through the assignment of an ordinal rating. Bridge ratings occur when a rater assigns two adjacent categories. Most statistical methodology necessitates the use of a single ordinal category. Consequently, bridge ratings often go unreported in clinical research studies. We propose three methodologies (Expanded, Mixture, and Collapsed) Bridge Category Models, to account for bridge ratings. We perform simulations to examine the impact of our approaches on detecting treatment effects, and comment on a real-world scenario of staging liver biopsies. Results indicate that if bridge ratings are not accounted for, disease staging models may exhibit significant bias and precision loss. All models worked well when they corresponded to the data generating mechanism.

pathology↗

bioRxiv2021

Mixed Effects Machine Learning Models for Colon Cancer Metastasis Prediction using Spatially Localized Immuno-Oncology Markers

Spatially resolved characterization of the transcriptome and proteome promises to provide further clarity on cancer pathogenesis and etiology, which may inform future clinical practice through classifier development for clinical outcomes. However, batch effects may potentially obscure the ability of machine learning methods to derive complex associations within spatial omics data. Profiling thirty-five stage three colon cancer patients using the GeoMX Digital Spatial Profiler, we found that mixed-effects machine learning (MEML) methods{dagger} may provide utility for overcoming significant batch effects to communicate key and complex disease associations from spatial information. These results point to further exploration and application of MEML methods within the spatial omics algorithm development life cycle for clinical deployment.

pathology↗

bioRxiv2020

Estimating latent positions of actors using Neural Networks in R with GCN4R

Network analysis methods are useful to better understand and contextualize relationships between entities. While statistical and machine learning prediction models generally assume independence between actors, network-based statistical methods for social network data allow for dyadic dependence between actors. While numerous methods have been developed for the R statistical software to analyze such data, deep learning methods have not been implemented in this language. Here, we introduce GCN4R, an R library for fitting graph neural networks on independent networks to aggregate actor covariate information to yield meaningful embeddings for a variety of network-based tasks (e.g. community detection, peer effects models, social influence). We provide an extensive overview of insights and methods utilized by the deep learning community on learning on social and biological networks, followed by a tutorial that demonstrates some of the capabilities of the GCN4R framework to make these methods more accessible to the R research community.

bioinformatics↗

bioRxiv2020

Activated iPSC-microglia from C9orf72 ALS/FTD patients exhibit endosomal-lysosomal dysfunction

While motor and cortical neurons are affected in C9orf72 ALS/FTD, it remains still largely unknown if and how non-neuronal cells induce or exacerbate neuronal damage. We generated C9orf72 ALS/FTD patient-derived induced pluripotent stem cells differentiated into microglia (iPSC-MG) and examined their intrinsic phenotypes. Similar to iPSC motor neurons, C9orf72 ALS/FTD iPSC-MG mono-cultures form G4C2 repeat RNA foci, exhibit reduced C9orf72 protein levels and generate dipeptide repeat proteins. Healthy control and C9orf72 iPSC-MG equivalently express microglial specific genes and display microglial functions including inflammatory cytokine release and phagocytosis of extracellular toxic cargos such as synthetic amyloid beta peptides and healthy human brain synaptoneurosomes. Select C9orf72 iPSC-MG patient lines show inability to efficiently remove phagocytosed contents, suggesting dysfunction of the endosomal-lysosomal pathways. Finally, RNA sequencing revealed overall transcriptional changes in diseased microglia yet no significant differentially expressed microglial-enriched genes. These minimal differences in cellular, molecular and functional characteristics of microglial mono-cultures suggest that a diseased microenvironment is associated with microglial activation and subsequent regulation of neuronal dysfunction.

neuroscience↗

bioRxiv2020

Investigation of Capsule-Inspired Neural Network Approaches for DNA Methylation

DNA methylation (DNAm) alterations have been heavily implicated in carcinogenesis and the pathophysiology of diseases through upstream regulation of gene expression. DNAm deep-learning approaches are able to capture features associated with aging, cell type, and disease progression, but lack incorporation of prior biological knowledge. Here, we present modular, user-friendly deep learning methodology and software, MethylCapsNet and MethylSPWNet, that group CpGs into biologically relevant capsules - such as gene promoter context, CpG island relationship, or user-defined groupings - and relate them to diagnostic and prognostic outcomes. We demonstrate these models utility on 3,897 individuals in the classification of central nervous system (CNS) tumors. MethylCapsNet and MethylSPWNet provide an opportunity to increase DNAm deep learning analyses interpretability by enabling a flexible organization of DNAm data into biologically relevant capsules.

bioinformatics↗

bioRxiv2020

Topological Feature Extraction and Visualization of Whole Slide Images using Graph Neural Networks

Whole-slide images (WSI) are digitized representations of thin sections of stained tissue from various patient sources (biopsy, resection, exfoliation, fluid) and often exceed 100,000 pixels in any given spatial dimension. Deep learning approaches to digital pathology typically extract information from sub-images (patches) and treat the sub-images as independent entities, ignoring contributing information from vital large-scale architectural relationships. Modeling approaches that can capture higher-order dependencies between neighborhoods of tissue patches have demonstrated the potential to improve predictive accuracy while capturing the most essential slide-level information for prognosis, diagnosis and integration with other omics modalities. Here, we review two promising methods for capturing macro and micro architecture of histology images, Graph Neural Networks, which contextualize patch level information from their neighbors through message passing, and Topological Data Analysis, which distills contextual information into its essential components. We introduce a modeling framework, WSI-GTFE that integrates these two approaches in order to identify and quantify key pathogenic information pathways. To demonstrate a simple use case, we utilize these topological methods to develop a tumor invasion score to stage colon cancer.

pathology↗

bioRxiv2020

A Large-Scale Internal Validation Study of Unsupervised Virtual Trichrome Staining Technologies on Non-alcoholic Steatohepatitis Liver Biopsies

Non-alcoholic steatohepatitis (NASH) is a fatty liver disease characterized by accumulation of fat in hepatocytes with concurrent inflammation and is associated with morbidity, cirrhosis and liver failure. After extraction of a liver core biopsy, tissue sections are stained with hematoxylin and eosin (H&E) to grade NASH activity, and stained with trichrome to stage fibrosis. Methods to computationally transform one stain into another on digital whole slide images (WSI) can lessen the need for additional physical staining besides H&E, reducing personnel, equipment, and time costs. Generative adversarial networks (GAN) have shown promise for virtual staining of tissue. We conducted a large-scale validation study of the viability of GANs for H&E to trichrome conversion on WSI (n=574). Pathologists were largely unable to distinguish real images from virtual/synthetic images given a set of twelve Turing Tests. We report high correlation between staging of real and virtual stains ({rho} = 0.86; 95% CI: 0.84-0.88). Stages assigned to both virtual and real stains correlated similarly with a number of clinical biomarkers and progression to End Stage Liver Disease (Hazard Ratio HR = 2.06, CI 95% 1.36-3.12, P < 0.001 for real stains; HR = 2.02, CI 95% 1.40-2.92, p < 0.001 for virtual stains). Our results demonstrate that virtual trichrome technologies may offer a software solution that can be employed in the clinical setting as a diagnostic decision aid.

pathology↗

bioRxiv2018

A pilot study of the effects of Mycoplasma ovipneumoniae exposure on domestic lamb growth and performance

Mycoplasma ovipneumoniae is a globally distributed pathogen that has been associated with pneumonia in both domestic and wild Caprinae. It is closely related to M. hyopneumoniae, a respiratory pathogen of swine that is associated with decreased growth rates of pigs as well as clinical respiratory disease. In order to assess the effects of M. ovipneumoniae on lamb performance, we generated a cohort of lambs free of M. ovipneumoniae by segregation of test negative ewes after lambing, then compared the growth and carcass quality traits of M. ovipneumoniae-free and -colonized lambs from weaning to harvest. Some signs of respiratory disease were observed during the feeding trial in both lamb groups, but the M. ovipneumoniae-exposed group included more affected lambs and higher average disease scores. At harvest, lungs of lambs in both groups showed few grossly visible lesions, although the M. ovipneumoniae-exposed group did exhibit increased microscopic lung lesions (P<0.05). In addition, M. ovipneumoniae exposed lambs produced lower average daily gains (P<0.05), and lower yield grade carcasses (P<0.05) compared to those of non-exposed lambs. The results demonstrated the feasibility of test and segregation for elimination of M. ovipneumoniae from groups of sheep and suggested that this pathogen may impair lamb growth and productivity even in the absence of overt respiratory disease.

microbiology↗

bioRxiv2018

OMA standalone: orthology inference among public and custom genomes and transcriptomes

Genomes and transcriptomes are now typically sequenced by individual labs, but analysing them often remains challenging. One essential step in many analyses lies in identifying orthologs--corresponding genes across multiple species--but this is far from trivial. The OMA (Orthologous MAtrix) database is a leading resource for identifying orthologs among publicly available, complete genomes. Here, we describe the OMA pipeline available as a standalone program for Linux and Mac. When run on a cluster, it has native support for the LSF, SGE, PBS Pro, and Slurm job schedulers and can scale up to thousands of parallel processes. Another key feature of OMA standalone is that users can combine their own data with existing public data by exporting genomes and pre-computed alignments from the OMA database, which currently contains over 2100 complete genomes. We compare OMA standalone to other methods in the context of phylogenetic tree inference, by inferring a phylogeny of the Lophotrochozoa, a challenging clade within the Protostomes. We also discuss other potential applications of OMA standalone, including identifying gene families having undergone duplications/losses in specific clades, and identifying potential drug targets in non-model organisms. OMA Standalone is available at http://omabrowser.org/standalone under the permissible open source Mozilla Public License Version 2.0.

bioinformatics↗

bioRxiv2017

A framework to identify modifier genes in patients with Phelan-McDermid syndrome

Phelan-McDermid syndrome (PMS) is characterized by a variety of clinical symptoms with heterogeneous degrees of severity, including intellectual disability, speech impairment, and autism spectrum disorders (ASD). It results from a deletion of the 22q13 locus that in most cases includes the SHANK3 gene. SHANK3 is considered a major gene for PMS, but the factors modulating the severity of the syndrome remain largely unknown. In this study, we investigated 85 PMS patients with different 22q13 rearrangements (78 deletions, 7 duplications). We first explored their clinical features and provide evidence for frequent corpus callosum abnormalities. We then mapped candidate genomic regions at the 22q13 locus associated with risk of clinical features, and suggest a second locus associated with absence of speech. Finally, in some cases, we identified additional rearrangements at loci associated with ASD, potentially modulating the severity of the syndrome. We also report the first SHANK3 deletion transmitted to five affected daughters by a mother without intellectual disability nor ASD, suggesting that some individuals could compensate for such mutations. In summary, we shed light on the genotype-phenotype relationship of PMS, a step towards the identification of compensatory mechanisms for a better prognosis and possibly treatments of patients with neurodevelopmental disorders.

genetics↗