Machine learning predicts individual cancer patient responses to therapeutic drugs with high accuracy
Ontology highlight
ABSTRACT:
PROVIDER: PRJNA449025 | ENA |
REPOSITORIES: ENA
Similar Datasets
Project description:Samples of primary tumors collected from 23 ovarian cancer patients
2018-04-07 | GSE112798 | GEO
Project description: Not available
| S-EPMC6219522 | biostudies-literature
Project description:Here we aim to test the notion that overcoming cancer cells’ acquired resistance to chemotherapy drugs may be achieved by administrating together with the main drug a molecule that is not toxic by itself but enhances the drug efficacy in killing the resistant cells. Finding such a supplementary molecule is a nontrivial task. Our algorithmic approach, named OVERcoming Drug Resistance by In Vitro cell Exposure (OVERDRIVE), involves exposing drug-sensitive cancer cells to increasing drug concentrations for several weeks followed by a comprehensive proteomic analysis of both the exposed and sensitive cells. Applying a machine learning tool Genome Enhancer to the proteomic data reveals the protein network accounting for the differences between the exposed and naïve cells and predicts molecules that may shift the exposed cell phenotypes to the sensitive one. Experimental validation of the top predicted nontoxic molecules confirmed that the combined treatment of the experimental drug LTCA2940 that targets redox pathways with identified supplementary molecule restored the sensitivity of LTCA2940-resistant cells. The OVERDRIVE approach may pave a way for a strategy to overcoming anticancer resistance to chemotherapy in clinic.
2026-06-08 | PXD071434 | Pride
Project description:Objectives: To predict response prior to anti-TNF treatment and comprehensively understand the mechanism how patients respond differently to anti-TNF treatment in rheumatoid arthritis (RA). Methods: Gene expression and/or DNA methylation profiling on PBMC, monocytes, and CD4+ T cells, from 80 RA patients before initiating either adalimumab (ADA) or etanercept (ETN) therapy was studied. After 6-month treatment response was evaluated according to the EULAR criteria of disease response. Differential expression and methylation analyses were performed to identify the response-associated transcriptional and epigenetic signatures. Machine learning models were built using these signatures by random forest algorithm to predict response prior to anti-TNF treatment and were further validated by a follow-up study. Results: Transcriptional signatures in ADA and ETN responders are divergent in PBMCs, and this phenomenon was reproduced in monocytes and CD4+ T cells. The genes upregulated in CD4+ T cells of ADA responders were enriched in the TNF signaling pathway, while very few pathways were differential in monocytes. Differential methylation positions (DMPs) of responders to ETN but not to ADA are majorly hypermethylated. The machine learning models to predict the response to ADA and ETN using differential genes reached overall accuracy of 85.9% and 79%, respectively. The models using DMPs reached overall accuracy of 84.7% and 88% for ADA and ETN, respectively. A follow-up study validated the high performance of these models. Conclusions: Machine learning models based on these molecular signatures could accurately predict response before ADA and ETN treatment, paving the path towards personalized anti-TNF treatment.
2020-08-24 | GSE138746 | GEO
Project description:Objectives: To predict response prior to anti-TNF treatment and comprehensively understand the mechanism how patients respond differently to anti-TNF treatment in rheumatoid arthritis (RA). Methods: Gene expression and/or DNA methylation profiling on PBMC, monocytes, and CD4+ T cells, from 80 RA patients before initiating either adalimumab (ADA) or etanercept (ETN) therapy was studied. After 6-month treatment response was evaluated according to the EULAR criteria of disease response. Differential expression and methylation analyses were performed to identify the response-associated transcriptional and epigenetic signatures. Machine learning models were built using these signatures by random forest algorithm to predict response prior to anti-TNF treatment and were further validated by a follow-up study. Results: Transcriptional signatures in ADA and ETN responders are divergent in PBMCs, and this phenomenon was reproduced in monocytes and CD4+ T cells. The genes upregulated in CD4+ T cells of ADA responders were enriched in the TNF signaling pathway, while very few pathways were differential in monocytes. Differential methylation positions (DMPs) of responders to ETN but not to ADA are majorly hypermethylated. The machine learning models to predict the response to ADA and ETN using differential genes reached overall accuracy of 85.9% and 79%, respectively. The models using DMPs reached overall accuracy of 84.7% and 88% for ADA and ETN, respectively. A follow-up study validated the high performance of these models. Conclusions: Machine learning models based on these molecular signatures could accurately predict response before ADA and ETN treatment, paving the path towards personalized anti-TNF treatment.
2020-08-24 | GSE138653 | GEO
Project description:Gene expression profiles were generated from 199 primary breast cancer patients. Samples 1-176 were used in another study, GEO Series GSE22820, and form the training data set in this study. Sample numbers 200-222 form a validation set. This data is used to model a machine learning classifier for Estrogen Receptor Status. RNA was isolated from 199 primary breast cancer patients. A machine learning classifier was built to predict ER status using only three gene features.
2013-01-01 | E-GEOD-29210 | biostudies-arrayexpress
Project description:Machine learning methods, particularly neural networks trained on large datasets, are transforming how scientists approach scientific discovery and experimental design. However, current state-of-the-art neural networks are limited by their uninterpretability: despite providing accurate predictions, they cannot describe how they arrived at their predictions. Here, using an ``interpretable-by-design'' approach, we present a neural network model that provides insights into RNA splicing, a fundamental process in the transfer of genomic information into functional biochemical products. Although we designed our model to emphasize interpretability, its predictive accuracy is on par with state-of-the-art models. To demonstrate the model's interpretability, we introduce a visualization that, for any given exon, allows us to trace and quantify the entire decision process from input sequence to output splicing prediction. Importantly, the model revealed novel components of the splicing logic, which we experimentally validated. This study highlights how interpretable machine learning can advance scientific discovery.
2022-10-01 | GSE200096 | GEO
Project description:Background: Adenomyosis uteri is a chronic gynecological condition frequently coexisting with endometriosis, presenting significant diagnostic challenges due to overlapping symptoms and limitations in imaging techniques. There is a pressing need for reliable, non-invasive biomarkers to enhance diagnostic accuracy and improve patient care. Objective: This pilot study investigated the diagnostic potential of serum and urine microRNA (miRNA) profiles for adenomyosis using machine learning approaches. Methods: Serum and urine samples were collected from 59 patients undergoing surgery for chronic pelvic pain at the Endometriosis Center, RWTH Aachen University Hospital. Among them, 7 had isolated adenomyosis, 34 had histologically confirmed endometriosis, and 18 served as negative controls. miRNA profiling was conducted via next-generation sequencing. A comprehensive feature selection pipeline—including variance filtering, univariate testing, mutual information, and recursive feature elimination—was used to reduce dimensionality. Classification models (Logistic Regression, Random Forest, Support Vector Machine, and Decision Tree) were trained with cross-validation and evaluated using accuracy, precision, recall, F1 score, and ROC-AUC.
2026-05-27 | GSE298298 | GEO
Project description:Machine-learning model based on glycosyltransferase expression predicts multiple cancer types, subtypes and survival probability
| PRJNA1070610 | ENA
Project description:Whereas current tools for diagnosing cancer focus on identifying markers in limited types of the disease, such as the PAM50 kit known primarily for subtyping breast cancer, our research reveals that 27 different cancers have a unique glycosyltransferase (GT) fingerprint that can be detected using a single test, with 93.7% accuracy in external validation. Four models were built on the expression patterns of 71 GTs that are universally influential in inducing the formation of cancer cells and giving them the impetus to metastasize. The first differentiates cancer tissue from normal tissue, the second distinguishes cancer types, and the third and fourth models differentiate between subtypes of cancer, in particular breast cancer and glioma. Data from the Cancer Genome Atlas database were used for deriving the models, while independent publicly available databases and newly collected patient samples were used for validation. Running these models against external databases showed just how powerful they are, producing results that were highly comparable to those achieved in internal testing. Furthermore, the breast cancer classifier differentiated between breast cancer subtypes with an average of 81% accuracy in external testing, far surpassing the industry-standard PAM50’s accuracy ratio of 43%. Meanwhile, a separate prognostic model designed to predict the probability of survival among glioma patients achieved high accuracy by zooming in on four GT genes that are strongly related to prognosis. Taken together, the models demonstrate the revolutionary potential of focusing on GT genes to simplify cancer diagnostics.
2024-02-03 | GSE254461 | GEO