Featured
-
-
Article
| Open Access
Mechanism-centric regulatory network identifies NME2 and MYC programs as markers of Enzalutamide resistance in CRPC
Heterogeneous response to Enzalutamide remains a critical issue in castration-resistant prostate cancer (CRPC). Here, the authors reconstruct a CRPC-specific mechanism-centric regulatory network to identify signatures of Enzalutamide response and predict patients at risk of Enzalutamide resistance.
- Sukanya Panja
- , Mihai Ioan Truica
- & Antonina Mitrofanova
-
Article
| Open Access
Quick model-based viscoelastic clot strength predictions from blood protein concentrations for cybermedical coagulation control
Available viscoelastic models of blood flow and blood coagulation are unsuited for a cybermedical input-output type of control system application. Here the authors present validated viscoelastic coagulation models that use quickly-measurable protein concentrations to forecast slow clot strength curves for future automation.
- Damon E. Ghetmiri
- , Alessia J. Venturi
- & Amor A. Menezes
-
Article
| Open Access
Prediction and stratification of longitudinal risk for chronic obstructive pulmonary disease across smoking behaviors
Many people who never smoke develop COPD. Here, the authors derive and validate the Socioeconomic and Environmental Risk Score (SERS) which captures cumulative exposure risks beyond tobacco smoking to predict and stratify risk of COPD.
- Yixuan He
- , David C. Qian
- & Chirag J. Patel
-
Article
| Open Access
Deep learning of cell spatial organizations identifies clinically relevant insights in tissue images
Cell spatial organization in tissue provides essential insights into diseases. Here, the authors show Ceograph, a graph convolutional network, for the analysis of pathology images to predict patient outcomes, highlighting cellular markers to guide personalized treatments and enhance biological understanding.
- Shidan Wang
- , Ruichen Rong
- & Guanghua Xiao
-
Article
| Open Access
Inflammation in the tumor-adjacent lung as a predictor of clinical outcome in lung adenocarcinoma
Lung adenocarcinoma is often curable when diagnosed at an early stage, but a subsection of patients will progress. Here, the authors use multi-omics profiling to show that gene expression data can predict clinical outcome.
- Igor Dolgalev
- , Hua Zhou
- & Aristotelis Tsirigos
-
Article
| Open Access
Quantifying portable genetic effects and improving cross-ancestry genetic prediction with GWAS summary statistics
Polygenic risk scores are used to improve risk prediction for common diseases but typically have reduced accuracy for individuals of non-European ancestry. Here, the authors present an approach that improves polygenic risk score performance in ancestrally diverse populations.
- Jiacheng Miao
- , Hanmin Guo
- & Qiongshi Lu
-
Article
| Open Access
Pharmacogenomics polygenic risk score for drug response prediction using PRS-PGx methods
To try to predict an individual’s drug response using genetic data, most studies have used traditional polygenic risk score (PRS) methods. Here, the authors develop a pharmacogenomics-specific PRS method, which can improve drug response prediction and patient stratification in pharmacogenomics studies.
- Song Zhai
- , Hong Zhang
- & Judong Shen
-
Article
| Open Access
Transferability of genetic loci and polygenic scores for cardiometabolic traits in British Pakistani and Bangladeshi individuals
Most genetic studies of disease have been done in European ancestry cohorts, and the relevance to other populations is not guaranteed. Here, the authors use data from 22,000 British South Asian individuals and find that the transferability of polygenic scores was high for lipids and blood pressure, and lower for BMI and coronary artery disease.
- Qin Qin Huang
- , Neneh Sallah
- & Karoline Kuchenbaecker
-
Article
| Open Access
Recurrent somatic mutations as predictors of immunotherapy response
Few genetic biomarkers are known for cancer immunotherapy. Here the authors identify recurrently-mutated genes and pathways associated with treatment response and develop a classifier using tumour whole exome sequencing and clinical features.
- Zoran Z. Gajic
- , Aditya Deshpande
- & Neville E. Sanjana
-
Article
| Open Access
Progression of type 1 diabetes from latency to symptomatic disease is predicted by distinct autoimmune trajectories
Presence of islet autoantibodies precedes the onset of type 1 diabetes but it does not predict whether and how fast symptomatic disease appears. Here authors present a model to predict and visualize progression to diabetes by using a large longitudinal data set on autoantibodies and clinical parameters as input.
- Bum Chul Kwon
- , Vibha Anand
- & Brigitte I. Frohnert
-
Article
| Open Access
Prediction of biomarkers and therapeutic combinations for anti-PD-1 immunotherapy using the global gene network association
A lot of cancer patients are not responsive to anti-PD1 therapy. Here, the authors develop a network approach to identify genes, pathways and potential therapeutic combinations and develop an MHC-I gene immunoscore associated with tumour response to anti-PD1.
- Chia-Chin Wu
- , Y. Alan Wang
- & P. Andrew Futreal
-
Article
| Open Access
Advancing diagnostic performance and clinical usability of neural networks via adversarial training and dual batch normalization
Unmasking the decision making process of machine learning models is essential for implementing diagnostic support systems in clinical practice. Here, the authors demonstrate that adversarially trained models can significantly enhance the usability of pathology detection as compared to their standard counterparts.
- Tianyu Han
- , Sven Nebelung
- & Daniel Truhn
-
Article
| Open Access
Machine learning analyses of antibody somatic mutations predict immunoglobulin light chain toxicity
Systemic light chain amyloidosis (AL) is caused by the production of toxic light chains and can be fatal, yet effective treatments are often not possible due to delayed diagnosis. Here the authors show that a machine learning platform analyzing light chain somatic mutations allows the prediction of light chain toxicity to serve as a possible tool for early diagnosis of AL.
- Maura Garofalo
- , Luca Piccoli
- & Andrea Cavalli
-
Article
| Open Access
Drug ranking using machine learning systematically predicts the efficacy of anti-cancer drugs
Artificial intelligence and machine learning promise to transform cancer therapies by accurately predicting the most appropriate drugs to treat individual patients. Here, the authors present an approach which uses omics data to produce ordered lists of drugs based on their effectiveness in decreasing cancer cell proliferation.
- Henry Gerdes
- , Pedro Casado
- & Pedro R. Cutillas
-
Article
| Open Access
Gastrointestinal microbiota composition predicts peripheral inflammatory state during treatment of human tuberculosis
Antibiotic therapy can lead to pathogen clearance, but also to alterations in the gut microbiota and systemic immune responses. Here, the authors analyze data from patients with tuberculosis and healthy subjects to show that pathogen clearance and gut microbiota alterations are independently associated with antibiotic-induced changes of the inflammatory response of active tuberculosis.
- Matthew F. Wipperman
- , Shakti K. Bhattarai
- & Vanni Bucci
-
Article
| Open Access
Artificial intelligence in sepsis early prediction and diagnosis using unstructured data in healthcare
Early prediction and diagnosis of sepsis, which is critical in reducing mortality, is challenging as many of its signs and symptoms are similar to other less critical conditions. Here, the authors develop an artificial intelligence algorithm which uses both structured data and unstructured clinical notes to predict sepsis.
- Kim Huat Goh
- , Le Wang
- & Gamaliel Yu Heng Tan
-
Article
| Open Access
Interplay between chromosomal alterations and gene mutations shapes the evolutionary trajectory of clonal hematopoiesis
Patients with solid cancers have high rates of clonal haematopoiesis associated with increased risk of secondary leukemias. Here, by using peripheral blood sequencing data from patients with solid non-hematologic cancer, the authors profile the landscape of mosaic chromosomal alterations and gene mutations, defining patients at high risk of leukemia progression.
- Teng Gao
- , Ryan Ptashkin
- & Elli Papaemmanuil
-
Article
| Open Access
Cross-site transportability of an explainable artificial intelligence model for acute kidney injury prediction
Artificial intelligence (AI) has demonstrated promise in predicting acutekidney injury (AKI), however, clinical adoption of these models requires interpretability and transportability across sites. Here, the authors develop an AKI prediction model and a measure for model transportability across six independent health systems.
- Xing Song
- , Alan S. L. Yu
- & Mei Liu
-
Article
| Open Access
MEPE loss-of-function variant associates with decreased bone mineral density and increased fracture risk
Bone mineral density (BMD) is associated with fracture risk and many genetic loci with small effect sizes have been discovered by genome-wide association studies (GWAS). Here, the authors discover a large-effect rare loss-of-function genetic variant for BMD in the MEPE gene in the Norwegian HUNT study which replicates in the UK Biobank.
- Ida Surakka
- , Lars G. Fritsche
- & Cristen J. Willer
-
Article
| Open Access
Repertoire-scale determination of class II MHC peptide binding via yeast display improves antigen prediction
Identifying peptides that can bind major histocompatibility complex II (MHC-II) is important for our understanding of T cell immunity and specificity. Here the authors present a yeast-display library screening approach that identifies more potential binders than various reported algorithms to help expand our understanding for antigen presentation.
- C. Garrett Rappazzo
- , Brooke D. Huisman
- & Michael E. Birnbaum
-
Article
| Open Access
Improving the accuracy of medical diagnosis with causal machine learning
In medical diagnosis a doctor aims to explain a patient’s symptoms by determining the diseases causing them, while existing diagnostic algorithms are purely associative. Here, the authors reformulate diagnosis as a counterfactual inference task and derive new counterfactual diagnostic algorithms.
- Jonathan G. Richens
- , Ciarán M. Lee
- & Saurabh Johri
-
Article
| Open Access
Explainable artificial intelligence model to predict acute critical illness from electronic health records
Acute critical illness is often preceded by deterioration of routinely measured clinical parameters, e.g., blood pressure and heart rate. Here, the authors develop an explainable artificial intelligence early warning score system for its early detection.
- Simon Meyer Lauritsen
- , Mads Kristensen
- & Bo Thiesson
-
Perspective
| Open Access
Causality matters in medical imaging
Scarcity of high-quality annotated data and mismatch between the development dataset and the target environment are two of the main challenges in developing predictive tools from medical imaging. In this Perspective, the authors show how causal reasoning can shed new light on these challenges.
- Daniel C. Castro
- , Ian Walker
- & Ben Glocker
-
Article
| Open Access
Proteome activity landscapes of tumor cell lines determine drug responses
Proteome activity has a major role in cancer progression and response to drugs. Here, the authors use comprehensive proteomic and phosphoproteomic data, in conjunction with drug-sensitivity screens, to generate a community resource consisting of landscapes of pathway and kinase activity across different cell lines
- Martin Frejno
- , Chen Meng
- & Bernhard Kuster
-
Article
| Open Access
Early triage of critically ill COVID-19 patients using deep learning
The sudden deterioration of patients with novel coronavirus disease 2019 (COVID-19) into critical illness is of major concern and early assessment would be vital. Here, the authors show that a deep learning-based survival model can predict the risk of COVID-19 patients developing critical illness based on clinical characteristics at admission.
- Wenhua Liang
- , Jianhua Yao
- & Jianxing He
-
Article
| Open Access
Efficient polygenic risk scores for biobank scale data by exploiting phenotypes from inferred relatives
Genetic data from large cohorts of unrelated individuals can be used to create polygenic risk scores, which could be used to predict individual risk of developing a specific disease. Here the authors show that smaller cohorts of related individuals can provide similarly powerful predictive ability.
- Buu Truong
- , Xuan Zhou
- & S. Hong Lee
-
Article
| Open Access
Ancestry deconvolution and partial polygenic score can improve susceptibility predictions in recently admixed individuals
Polygenic scores are believed to hold future promise for trait prediction and personalized medicine, but are sensitive to demographic history. Here, Marnetto et al. develop partial polygenic scores supplemented with local ancestry deconvolution which improves prediction accuracy into recently admixed European populations.
- Davide Marnetto
- , Katri Pärna
- & Luca Pagani
-
Article
| Open Access
Estimating heritability and genetic correlations from large health datasets in the absence of genetic data
Disease heritability and genetic correlations between traits depend on genetics, the environment and their interaction. Here, Jia et al. compute disease prevalence curves and disease embeddings from electronic health records and impute heritability for hundreds of diseases and genetic correlations for thousands of disease pairs.
- Gengjie Jia
- , Yu Li
- & Andrey Rzhetsky
-
Article
| Open Access
Dissection of gene expression datasets into clinically relevant interaction signatures via high-dimensional correlation maximization
Identification of clinically relevant gene expression signatures for cancer stratification remains challenging. Here, the authors introduce a flexible nonlinear signal superposition model that enables dissection of large gene expression data sets into signatures and extraction of gene interactions.
- Michael Grau
- , Georg Lenz
- & Peter Lenz
-
Article
| Open Access
Predicting bacterial infection outcomes using single cell RNA-sequencing analysis of human immune cells
Complex interactions between different host immune cell types can determine the outcome of pathogen infections. Here, Avraham and colleagues present a deconvolution algorithm that uses single-cell RNA and bulk RNA sequencing measurements of pathogen-infected cells to predict disease risk outcomes.
- Noa Bossel Ben-Moshe
- , Shelly Hen-Avivi
- & Roi Avraham
-
Article
| Open Access
Integrating biomedical research and electronic health records to create knowledge-based biologically meaningful machine-readable embeddings
The Scalable Precision Medicine Oriented Knowledge Engine (SPOKE) is a heterogeneous knowledge network that integrates information from 29 public databases. Here, Nelson et al. extend SPOKE to embed clinical data from electronic health records to create medically meaningful barcodes for each medical variable.
- Charlotte A. Nelson
- , Atul J. Butte
- & Sergio E. Baranzini
-
Article
| Open Access
Conserved transcriptomic profile between mouse and human colitis allows unsupervised patient stratification
Clinical and molecular heterogeneity of ulcerative colitis presents unresolved challenges to identify predictive biomarkers of response to therapies. Here, the authors combine mouse colitis time course with patient biopsy transcriptomes, achieving unsupervised clustering of UC patients correlating with therapeutic outcomes in independent data sets.
- Paulo Czarnewski
- , Sara M. Parigi
- & Eduardo J. Villablanca
-
Article
| Open Access
An additive Gaussian process regression model for interpretable non-parametric analysis of longitudinal data
Longitudinal data are common in biomedical research, but their analysis is often challenging. Here, the authors present an additive Gaussian process regression model specifically designed for statistical analysis of longitudinal experimental data.
- Lu Cheng
- , Siddharth Ramchandran
- & Harri Lähdesmäki
-
Article
| Open Access
Metabolite changes in blood predict the onset of tuberculosis
The tuberculosis pandemic requires new methods for diagnosing and containing infections prior to active disease. Here, the authors performed a multi-site observational study within sub-Saharan Africa and present serum and plasma metabolic signatures that can predict the onset of active TB with a high degree of sensitivity and specificity.
- January Weiner 3rd
- , Jeroen Maertzdorf
- & Sarah Zalwango
-
Article
| Open Access
Genetic signature to provide robust risk assessment of psoriatic arthritis development in psoriasis patients
Approximately 30% of psoriasis patients develop psoriatic arthritis (PsA) and early diagnosis is crucial for the management of PsA. Here, Patrick et al. develop a computational pipeline involving statistical and machine-learning methods that can assess the risk of progression to PsA based on genetic markers.
- Matthew T. Patrick
- , Philip E. Stuart
- & Lam C. Tsoi
-
Article
| Open Access
Pan-cancer deconvolution of tumour composition using DNA methylation
Determining the extent of immune cell infiltration into solid tumours is essential for adequate therapeutic response. Here the authors develop a DNA methylation-based approach for tumour cell fraction deconvolution and analyse tumour composition and genomics across a wide spectrum of solid cancers.
- Ankur Chakravarthy
- , Andrew Furness
- & Tim R. Fenton
-
Article
| Open Access
Predicting treatment benefit in multiple myeloma through simulation of alternative treatment effects
Selection of the right cancer treatment is still a challenge. Here, the authors introduce a framework to analyze treatment benefits, using the idea that patients with similar genetic tumor profiles receiving different treatments can be used to model their responses to the alternative treatment.
- Joske Ubels
- , Pieter Sonneveld
- & Jeroen de Ridder
-
Article
| Open Access
Multi-omics monitoring of drug response in rheumatoid arthritis in pursuit of molecular remission
Little information is available on molecular changes in response to treatment of rheumatoid arthritis (RA). Here the authors report a multi-omics study collecting patients' transcriptome, proteome, and immunophenotype data to help understand the impact of drug treatments on RA molecular phenotypes.
- Shinya Tasaki
- , Katsuya Suzuki
- & Tsutomu Takeuchi
-
Article
| Open Access
Network-based approach to prediction and population-based validation of in silico drug repurposing
Repurposing approved drugs could accelerate treatment options for various diseases. Here, the authors use network proximity of disease gene products and drug targets in the human protein interactome to identify drug-disease associations for cardiovascular disease, and validate these using longitudinal healthcare data.
- Feixiong Cheng
- , Rishi J. Desai
- & Joseph Loscalzo
-
Article
| Open Access
Improving genetic prediction by leveraging genetic correlations among human diseases and traits
Genetic prediction of complex traits so far has limited accuracy because of insufficient understanding of the genetic risk. Here, Maier et al. develop an improved method for trait prediction that makes use of genetic correlations between traits and apply it to summary statistics of psychiatric diseases.
- Robert M. Maier
- , Zhihong Zhu
- & Matthew R. Robinson
-
Article
| Open Access
Perturbation-response genes reveal signaling footprints in cancer gene expression
Deregulation of signalling is responsible for many cancer phenotypes. Leveraging available perturbation data, here the authors assess large-scale pathway activity patterns based on consensus downstream readout genes, enabling accurate prediction of the effects of mutations and small molecules.
- Michael Schubert
- , Bertram Klinger
- & Julio Saez-Rodriguez
-
Article
| Open Access
An extracellular matrix-related prognostic and predictive indicator for early-stage non-small cell lung cancer
Prognosis and prediction of adjuvant chemotherapy response in non-small cell lung cancer can have significant clinical impact. Here, the authors show that differential expression of a 29 extracellular matrix gene indicator, EPPI, can predict patient outcome.
- Su Bin LIM
- , Swee Jin TAN
- & Chwee Teck LIM
-
Article
| Open Access
Gene isoforms as expression-based biomarkers predictive of drug response in vitro
Altered mRNA splicing features in many cancers, but it has not been linked to drug response. Here, with their meta-analytic framework, the authors analyse pharmacogenomic data to identify isoform-based biomarkers predictive of in vitro drug response, and show them to frequently be strong predictors.
- Zhaleh Safikhani
- , Petr Smirnov
- & Benjamin Haibe-Kains
-
Article
| Open Access
A molecular portrait of microsatellite instability across multiple cancers
Some cancers with DNA mismatch repair deficiency display microsatellite instability. Here the authors analyse twenty three cancer types at the exome and whole-genome level, and identify loci with recurrent microsatellite instability that could be used to identify patients who would benefit from immunotherapy.
- Isidro Cortes-Ciriano
- , Sejoon Lee
- & Peter J. Park
-
Article
| Open Access
In silico Pathway Activation Network Decomposition Analysis (iPANDA) as a method for biomarker development
Pathway analysis aids interpretation of large-scale gene expression data, but existing algorithms fall short of providing robust pathway identification. The method introduced here includes coexpression analysis and gene importance estimation to robustly identify relevant pathways and biomarkers for patient stratification.
- Ivan V. Ozerov
- , Ksenia V. Lezhnina
- & Alex Zhavoronkov
-
Article
| Open Access
Tissue-specific and convergent metabolic transformation of cancer correlates with metastatic potential and patient survival
Cancer cells reprogramme their metabolism with unclear clinical implications. Here, the authors analyse the expression of metabolic genes across 20 types of solid cancers and find that clinical aggressiveness, poor survival and metastasis are associated with the deregulation of mitochondrial metabolism.
- Edoardo Gaude
- & Christian Frezza
-
Article
| Open Access
Crowdsourced assessment of common genetic contribution to predicting anti-TNF treatment response in rheumatoid arthritis
Rheumatoid arthritis patients respond differently to anti-TNF treatment. Using community-based challenge, the authors show that currently available data does not reveal meaningful genetic predictors of response to anti-TNF therapy, thus confirming clinical observations.
- Solveig K. Sieberts
- , Fan Zhu
- & Lara M. Mangravite
-
Article
| Open Access
Predicting non-small cell lung cancer prognosis by fully automated microscopic pathology image features
Diagnosis of lung cancer through manual histopathology evaluation is insufficient to predict patient survival. Here, the authors use computerized image processing to identify diagnostically relevant image features and use these features to distinguish lung cancer patients with different prognoses.
- Kun-Hsing Yu
- , Ce Zhang
- & Michael Snyder
Data-driven identification of predictive risk biomarkers for subgroups of osteoarthritis using interpretable machine learning