Cancer Subtype Classification Research Articles

MoBRCA-net: a breast cancer subtype classification framework based on multi-omics attention neural networks

BackgroundBreast cancer is a highly heterogeneous disease that comprises multiple biological components. Owing its diversity, patients have different prognostic outcomes; hence, early diagnosis and accurate subtype prediction are critical for treatment. Standardized breast cancer subtyping systems, mainly based on single-omics datasets, have been developed to ensure proper treatment in a systematic manner. Recently, multi-omics data integration has attracted attention to provide a comprehensive view of patients but poses a challenge due to the high dimensionality. In recent years, deep learning-based approaches have been proposed, but they still present several limitations.ResultsIn this study, we describe moBRCA-net, an interpretable deep learning-based breast cancer subtype classification framework that uses multi-omics datasets. Three omics datasets comprising gene expression, DNA methylation and microRNA expression data were integrated while considering the biological relationships among them, and a self-attention module was applied to each omics dataset to capture the relative importance of each feature. The features were then transformed to new representations considering the respective learned importance, allowing moBRCA-net to predict the subtype.ConclusionsExperimental results confirmed that moBRCA-net has a significantly enhanced performance compared with other methods, and the effectiveness of multi-omics integration and omics-level attention were identified. moBRCA-net is publicly available at https://github.com/cbi-bioinfo/moBRCA-net.

Lung cancer subtyping from gene expression data using general and enhanced Fuzzy min–max neural networks

AbstractIn this article, we address the problem of lung cancer diagnosis from gene expression data, which is now recognized as an effective means for early treatment and prevention of cancer. Specifically, we employ Fuzzy min–max (FMM) classifier for the task which is a well‐known neuro‐fuzzy neural network. The idea is to take advantage of the fuzzy class definitions whose boundaries are set using the min–max hyperbox constructed for each class. We implement two advanced FMM neural network architectures: general Fuzzy min–max (GFMM) and enhanced Fuzzy min–max (EFMM) for the classification of lung cancer subtypes from gene expression data. The advantage of GFMM is that it involves very simple operations for hyperbox manipulation, and can handle both labeled and unlabeled data. On the other hand, EFMM proposes three heuristic rules related to hyperbox expansion, contraction and the overlap test, which enhances the learning algorithm. We perform the classification of gene expression data using these two models, and then we analyze the performance by visualizing the hyperboxes obtained after training, and compare the accuracies of these classifiers with the state of the art. Least absolute shrinkage and selection operator (LASSO) is used for selecting the informative genes from the high‐dimensional gene expression data. From the empirical results, we observe that GFMM with LASSO gives the best performance of all, with validation accuracy of 98.04% and cross‐validation accuracy of 94.06%.

SpongEffects: ceRNA modules offer patient-specific insights into the miRNA regulatory landscape.

Cancer is one of the leading causes of death worldwide. Despite significant improvements in prevention and treatment, mortality remains high for many cancer types. Hence, innovative methods that use molecular data to stratify patients and identify biomarkers are needed. Promising biomarkers can also be inferred from competing endogenous RNA (ceRNA) networks that capture the gene-miRNA gene regulatory landscape. Thus far, the role of these biomarkers could only be studied globally but not in a sample-specific manner. To mitigate this, we introduce spongEffects, a novel method that infers subnetworks (or modules) from ceRNA networks and calculates patient- or sample-specific scores related to their regulatory activity. We show how spongEffects can be used for downstream interpretation and machine learning tasks such as tumor classification and for identifying subtype-specific regulatory interactions. In a concrete example of breast cancer subtype classification, we prioritize modules impacting the biology of the different subtypes. In summary, spongEffects prioritizes ceRNA modules as biomarkers and offers insights into the miRNA regulatory landscape. Notably, these module scores can be inferred from gene expression data alone and can thus be applied to cohorts where miRNA expression information is lacking. https://bioconductor.org/packages/devel/bioc/html/SPONGE.html. are available at Bioinformatics online.

Using feature selection and Bayesian network identify cancer subtypes based on proteomic data

The Cancer Proteome Atlas (TCPA) project collects reverse-phase protein arrays (RPPA)-based proteome datasets from nearly 8000 samples across 32 cancer types. This study aims to investigate the pan-cancer proteome signature and identify cancer subtypes of glioma, kidney cancer, and lung cancer based on TCPA data. We first visualized the tumor clustering models using t-distributed stochastic neighbour embedding (t-SNE) and bi-clustering heatmap. Then, three feature selection methods (pyHSICLasso, XGBoost, and Random Forest) were performed to select protein features for classifying cancer subtypes in training dataset, and the LibSVM algorithm was empolyed to test classification accuracy in the validation dataset. Clustering analysis revealed that different kinds of tumors have relatively distinct proteomic profiling based on tissue or origin. We identified 20, 10, and 20 protein features with the highest accuracies in classifying subtypes of glioma, kidney cancer, and lung cancer, respectively. The predictive abilities of the selected proteins were confirmed by receiving operating characteristic (ROC) analysis. Finally, the Bayesian network was utilized to explore the protein biomarkers that have direct causal relationships with cancer subtypes. Overall, we highlight the theoretical and technical applications of machine learning based feature selection approaches in the analysis of high-throughput biological data, particularly for cancer biomarker research. SignificanceFunctional proteomics is a powerful approach for characterizing cell signaling pathways and understanding their phenotypic effects on cancer development. The TCPA database provides a platform to explore and analyze TCGA pan-cancer RPPA-based protein expression. With the advent of the RPPA technology, the availability of high-throughput data in TCPA platform has made it possible to use machine learning methods to identify protein biomarkers and further differentiate subtypes of cancer based on proteomic data. In this study, we highlight the role of feature selection and Bayesian network in discovery protein biomarker for classifying cancer subtypes based on functional proteomic data. The application of machine learning methods in the analysis of high-throughput biological data, particularly for cancer biomarker researches, which have potential clinical values in developing individualized treatment strategies.

FDTrans: Frequency Domain Transformer Model for predicting subtypes of lung cancer using multimodal data

Background and purpose:Accurate identification of lung cancer subtypes in medical images is of great significance for the diagnosis and treatment of lung cancer. Despite substantial progress in existing methods, they remain challenging due to limited annotated datasets, large intra-class differences, and high inter-class similarities. Methods:To address these challenges, we propose a Frequency Domain Transformer Model (FDTrans) to identify patients’ lung cancer subtypes using the TCGA lung cancer dataset. We add a pre-processing process to transfer histopathological images to the frequency domain using a block-based discrete cosine transform and design a coordinate Coordinate-Spatial Attention Module (CSAM) to obtain critical detail information by reassigning weights to the location information and channel information of different frequency vectors. Then, a Cross-Domain Transformer Block (CDTB) is designed for Y, Cb, and Cr channel features, capturing the long-term dependencies and global contextual connections between different component features. At the same time, feature extraction is performed on the genomic data to obtain specific features. Finally, the image branch and the gene branch are fused, and the classification result is output through the fully connected layer. Results:In 10-fold cross-validation, the method achieves an AUC of 93.16% and overall accuracy of 92.33%, which is better than similar current lung cancer subtypes classification detection methods. Conclusion:This method can help physicians diagnose the subtypes classification of lung cancer in patients and can benefit from both spatial and frequency domain information.

CustOmics: A versatile deep-learning based strategy for multi-omics integration

The availability of patient cohorts with several types of omics data opens new perspectives for exploring the disease's underlying biological processes and developing predictive models. It also comes with new challenges in computational biology in terms of integrating high-dimensional and heterogeneous data in a fashion that captures the interrelationships between multiple genes and their functions. Deep learning methods offer promising perspectives for integrating multi-omics data. In this paper, we review the existing integration strategies based on autoencoders and propose a new customizable one whose principle relies on a two-phase approach. In the first phase, we adapt the training to each data source independently before learning cross-modality interactions in the second phase. By taking into account each source's singularity, we show that this approach succeeds at taking advantage of all the sources more efficiently than other strategies. Moreover, by adapting our architecture to the computation of Shapley additive explanations, our model can provide interpretable results in a multi-source setting. Using multiple omics sources from different TCGA cohorts, we demonstrate the performance of the proposed method for cancer on test cases for several tasks, such as the classification of tumor types and breast cancer subtypes, as well as survival outcome prediction. We show through our experiments the great performances of our architecture on seven different datasets with various sizes and provide some interpretations of the results obtained. Our code is available on (https://github.com/HakimBenkirane/CustOmics).

A classification method of gastric cancer subtype based on residual graph convolution network.

Background: Clinical diagnosis and treatment of tumors are greatly complicated by their heterogeneity, and the subtype classification of cancer frequently plays a significant role in the subsequent treatment of tumors. Presently, the majority of studies rely far too heavily on gene expression data, omitting the enormous power of multi-omics fusion data and the potential for patient similarities. Method: In this study, we created a gastric cancer subtype classification model called RRGCN based on residual graph convolutional network (GCN) using multi-omics fusion data and patient similarity network. Given the multi-omics data's high dimensionality, we built an artificial neural network Autoencoder (AE) to reduce the dimensionality of the data and extract hidden layer features. The model is then built using the feature data. In addition, we computed the correlation between patients using the Pearson correlation coefficient, and this relationship between patients forms the edge of the graph structure. Four graph convolutional network layers and two residual networks with skip connections make up RRGCN, which reduces the amount of information lost during transmission between layers and prevents model degradation. Results: The results show that RRGCN significantly outperforms other classification methods with an accuracy as high as 0.87 when compared to four other traditional machine learning methods and deep learning models. Conclusion: In terms of subtype classification, RRGCN excels in all areas and has the potential to offer fresh perspectives on disease mechanisms and disease progression. It has the potential to be used for a broader range of disorders and to aid in clinical diagnosis.

An Intelligent Decision Support System for Lung Cancer Diagnosis

Lung cancer is the leading cause of cancer-related death around the globe. The treatment and survival rates among lung cancer patients are significantly impacted by early diagnosis. Most diagnostic techniques can identify and classify only one type of lung cancer. It is crucial to close this gap with a system that detects all lung cancer types. This paper proposes an intelligent decision support system for this purpose. This system aims to support the quick and early detection and classification of all lung cancer types and subtypes to improve treatment and save lives. Its algorithm uses a Convolutional Neural Network (CNN) tool to perform deep learning and a Random Forest Algorithm (RFA) to help classify the type of cancer present using several extracted features, including histograms and energy. Numerous simulation experiments were conducted on MATLAB, evidencing that this system achieves 98.7% accuracy and over 98% precision and recall. A comparative assessment assessing accuracy, recall, precision, specificity, and F-score between the proposed algorithm and works from the literature shows that the proposed system in this study outperforms existing methods in all considered metrics. This study found that using CNNs and RFAs is highly effective in detecting lung cancer, given the high accuracy, precision, and recall results. These results lead us to believe that bringing this kind of technology to doctors diagnosing lung cancer is critical.

Classification of breast cancer subtypes: A study based on representative genes

Breast cancer is the second most common cancer type and is the leading cause of cancer-related deaths worldwide. Since it is a heterogeneous disease, subtyping breast cancer plays an important role in performing a specific treatment. In this work, we propose an evaluation framework that uses different machine learning techniques for a broader analysis of the PAM50 list in the classification of breast cancer subtypes. The experiments show that the best method to be used in the classification of breast cancer subtypes is the SVM with linear kernel, which presented an F1 score of 0.98 for the Basal subtype and 0.90 for the Her 2 subtype, the two subtypes with worse prognosis, respectively. We also presented a gene analysis for the classification methods using SHAP values, where we found which genes are important for the classification of each subtype.

Efficient Breast Cancer Classification Network with Dual Squeeze and Excitation in Histopathological Images.

Medical image analysis methods for mammograms, ultrasound, and magnetic resonance imaging (MRI) cannot provide the underline features on the cellular level to understand the cancer microenvironment which makes them unsuitable for breast cancer subtype classification study. In this paper, we propose a convolutional neural network (CNN)-based breast cancer classification method for hematoxylin and eosin (H&E) whole slide images (WSIs). The proposed method incorporates fused mobile inverted bottleneck convolutions (FMB-Conv) and mobile inverted bottleneck convolutions (MBConv) with a dual squeeze and excitation (DSE) network to accurately classify breast cancer tissue into binary (benign and malignant) and eight subtypes using histopathology images. For that, a pre-trained EfficientNetV2 network is used as a backbone with a modified DSE block that combines the spatial and channel-wise squeeze and excitation layers to highlight important low-level and high-level abstract features. Our method outperformed ResNet101, InceptionResNetV2, and EfficientNetV2 networks on the publicly available BreakHis dataset for the binary and multi-class breast cancer classification in terms of precision, recall, and F1-score on multiple magnification levels.

Identification of necroptosis subtypes and development of necroptosis-related risk score model for in ovarian cancer.

Background: ith the ongoing development of targeted therapy, non-apoptotic cell death, including necroptosis, has become a popular topic in the field of prevention and treatment. The purpose of this study was to explore the effect of necroptosis-related genes (NRGs) on the classification of ovarian cancer (OV) subtypes and to develop a necroptosis-related risk score (NRRS) classification system. Methods: 74 NRGs were obtained from the published studies, and univariate COX regression analysis was carried out between them and OV survival. Consensus clustering analysis was performed on OV samples according to the expression of NRGs related to prognosis. Furthermore, the NRRS model was developed by combining Weighted Gene Co-Expression Network Analysis (WGCNA) with least absolute shrinkage and selection operator (Lasso)-penalized Cox regression and multivariate Cox regression analysis. And the decision tree model was constructed based on the principle of random forest screening factors principle. Results: According to the post-related NRGs, OV was divided into two necroptosis subtypes. Compared with Cluster 1 (C1), the overall survival (OS) of Cluster 2 (C2) was significantly shorter, stromal score and immune score, the infiltration level of tumor associated immune cells and the expression of 20 immune checkpoints were significantly higher. WGCNA identified the blue module most related to necroptosis subtype, and 12 genes in the module were used to construct NRRS. NRRS was an independent prognostic variable of OV. The OS of samples with lower NRRS was significantly longer, and tumor mutation burden and homologous recombination defect were more obvious. Conclusion: This study showed that necroptosis plays an important role in the classification, prognosis, immune infiltration and biological characteristics of OV subtypes. The evaluation of tumor necroptosis may provide a new perspective for OV treatment.

Research progression of gene fusion detection technology based on next generation sequencing in tumor companion diagnostics

Gene fusion is one of the mechanisms that promote tumor development. It is also an important cause for the poor prognosis of patients. The detection of gene fusion is crucial for the recognition of tumor biomarker, cancer subtype classification, and clinical medication guidance. Appropriate methods can help the early diagnosis and avoid ineffective medication. Traditional tests include fluorescence in situ hybridization (FISH), immunohistochemistry (IHC), reverse transcription of PCR (RT-PCR), and next generation sequencing (NGS). The next generation sequencing (NGS) mainly includes: whole genome sequencing (WGS), whole transcriptome sequencing (WTS) and target sequencing (hybridization capture method/amplicon method). In clinical concomitant diagnostic applications, some factors such as operability, time/money costs, and the level of expertise required for data analysis should be considered. This article concludes with a discussion of the technical principles of different detection methods and advantages/limitations. Meanwhile, it provides reference opinions for the detection methods of gene fusion.

A Unified Multi-Phase CT Synthesis and Classification Framework for Kidney Cancer Diagnosis With Incomplete Data.

Multi-phase computed tomography (CT) is widely adopted for the diagnosis of kidney cancer due to the complementary information among phases. However, the complete set of multi-phase CT is often not available in practical clinical applications. In recent years, there have been some studies to generate the missing modality image from the available data. Nevertheless, the generated images are not guaranteed to be effective for the diagnosis task. In this paper, we propose a unified framework for kidney cancer diagnosis with incomplete multi-phase CT, which simultaneously recovers missing CT images and classifies cancer subtypes using the completed set of images. The advantage of our framework is that it encourages a synthesis model to explicitly learn to generate missing CT phases that are helpful for classifying cancer subtypes. We further incorporate lesion segmentation network into our framework to exploit lesion-level features for effective cancer classification in the whole CT volumes. The proposed framework is based on fully 3D convolutional neural networks to jointly optimize both synthesis and classification of 3D CT volumes. Extensive experiments on both in-house and external datasets demonstrate the effectiveness of our framework for the diagnosis with incomplete data compared with state-of-the-art baselines. In particular, cancer subtype classification using the completed CT data by our method achieves higher performance than the classification using the given incomplete data.

Data and knowledge co-driving for cancer subtype classification on multi-scale histopathological slides

Artificial intelligence-enabled histopathological data analysis has become a valuable assistant to the pathologist. However, existing models lack representation and inference abilities compared with those of pathologists, especially in cancer subtype diagnosis, which is unconvincing in clinical practice. For instance, pathologists typically observe the lesions of a slide from global to local, and then can give a diagnosis based on their knowledge and experience. In this paper, we propose a Data and Knowledge Co-driving (D&K) model to replicate the process of cancer subtype classification on a histopathological slide like a pathologist. Specifically, in the data-driven module, the bagging mechanism in ensemble learning is leveraged to integrate the histological features from various bags extracted by the embedding representation unit. Furthermore, a knowledge-driven module is established based on the Gestalt principle in psychology to build the three-dimensional (3D) expert knowledge space and map histological features into this space for metric. Then, the diagnosis can be made according to the Euclidean distance between them. Extensive experimental results on both public and in-house datasets demonstrate that the D&K model has a high performance and credible results compared with the state-of-the-art methods for diagnosing histopathological subtypes. Code: https://github.com/Dennis-YB/Data-and-Knowledge-Co-driving-for-Cancer-Subtypes-Classification.

Classification of breast cancer subtypes based on RNA profiling and immunohistochemical methods: clinical and biological aspects: A review

Transcriptome analysis provided a tool to identify expression subtypes of breast cancer (BC). A significant part of BCs are carcinomas that differ in the expression of luminal mammary ductal epithelium markers and estrogen signaling cascade genes (luminal subtypes A and B). Another group of BCs is characterized by the expression of ductal basal lining genes. Another subtype of BC has genes typically expressed in HER2-induced tumors. Immunohistochemical (IHC) examination is reliable in identifying the tumor subtype. Tumors with high IHC-expression of estrogen (ER) and progesterone (PR) receptors with no signs of HER2 gene activation and low proliferative activity should be referred to as luminal type A. The absence of ER, PR, and HER2 expression should be considered a sign of basal tumor subtype. However, the IHC classification cannot reliably distinguish HER2-positive tumors and HER2-enriched subtypes, which do not reflect the biological features of some BCs. For instance, a significant number of ER-positive breast cancer patients included in the MONALEESA ribociclib clinical study had HER2-subtype transcriptional signatures in the absence of HER2 receptor expression, and these were the ones who demonstrated a pronounced response to treatment.

Explainable Multi-Class Classification Based on Integrative Feature Selection for Breast Cancer Subtyping

Breast cancer subtype classification is a multi-class classification problem that can be handled using computational methods. Three main challenges need to be addressed. Consider first the high dimensionality of the available datasets relative to the extremely small number of instances. Second, the integration of different levels of data makes the dimensionality problem even more challenging. The third challenging issue is the ability to explain the predictions provided by a machine learning model. Recently, several deep learning models have been proposed for feature extraction and classification. However, due to the small size of the datasets, they were unable to achieve satisfactory results, particularly in multi-class classification. Aside from that, explaining the impact of features on classification has not been addressed in previous works. To cope with these problems, we propose a multi-stage feature selection (FS) framework with two data integration schemes. Using multi-omics data, four machine learning models, namely support vector machines, random forest, extra trees, and XGBoost, were investigated at each level. The SHAP framework was used to explain how specific features influenced classification. Experimental results demonstrated that ensemble models with early integration and two stage feature selection improved results compared to baseline experiments and to state-of-the art methods. Furthermore, more explanations regarding the implications of the main relevant features in the predictions are provided, which could serve as a baseline for future biological investigations.

A novel Multi-Level feature selection method for radiomics

Radiomics is characterized by high-dimension and high redundancy. The existing Lasso-based feature selection does not consider features that are weakly correlated with the classification results, which will have a certain impact on the quality of feature subset. A multi-level feature selection algorithm based on Lasso coefficient threshold (Coe-Thr-Lasso) was proposed. Firstly, t-test and variance were used to remove the features that had little correlation with the classification results. Secondly, the proposed algorithm was used to remove features with redundancy and weak correlation of classification results. Three machine learning algorithms, including Logistic regression (LR), random forest (RF) and support vector machine (SVM), were verify the performance of the proposed algorithm on the non-small cell lung cancer subtype classification dataset. When modeling based on the feature subset generated by the proposed method, the proposed method achieved the best classification performance compared with other publication methods. Therefore, Coe-Thr-Lasso algorithm can effectively remove redundant and irrelevant features in radiomics, so as to improve the quality of feature subset and the ability of model generalization.

Deep Learning-Based Classification of Uterine Cervical and Endometrial Cancer Subtypes from Whole-Slide Histopathology Images.

Uterine cervical and endometrial cancers have different subtypes with different clinical outcomes. Therefore, cancer subtyping is essential for proper treatment decisions. Furthermore, an endometrial and endocervical origin for an adenocarcinoma should also be distinguished. Although the discrimination can be helped with various immunohistochemical markers, there is no definitive marker. Therefore, we tested the feasibility of deep learning (DL)-based classification for the subtypes of cervical and endometrial cancers and the site of origin of adenocarcinomas from whole slide images (WSIs) of tissue slides. WSIs were split into 360 × 360-pixel image patches at 20× magnification for classification. Then, the average of patch classification results was used for the final classification. The area under the receiver operating characteristic curves (AUROCs) for the cervical and endometrial cancer classifiers were 0.977 and 0.944, respectively. The classifier for the origin of an adenocarcinoma yielded an AUROC of 0.939. These results clearly demonstrated the feasibility of DL-based classifiers for the discrimination of cancers from the cervix and uterus. We expect that the performance of the classifiers will be much enhanced with an accumulation of WSI data. Then, the information from the classifiers can be integrated with other data for more precise discrimination of cervical and endometrial cancers.

The ubiquitin ligase RNF2 stabilizes ERα and modulates breast cancer progression

Estrogen receptor α (ERα) is the most common clinical marker used for breast cancer prognosis and the classification of breast cancer subtypes. Clinically, patients with estrogen receptor-positive breast cancer can receive endocrine therapy. However, resistance to endocrine therapy has become an urgent clinical problem. A large number of previous studies have proven that posttranslational modification of the estrogen receptor is significantly related to endocrine therapy resistance. RNF2 is a member of the RING finger protein family that functions as an E3 ubiquitin ligase. Several studies have clarified that RNF2 is a critical regulator of ERα transcriptional regulation. In our current study, we identified RNF2 as an important posttranslational modification regulator of the estrogen receptor. RNF2 depletion inhibited breast cancer cell progression and ERα signaling activity. TCGA data analysis indicated that RNF2 was elevated in breast malignancies, while RNF2 depletion could drastically inhibit estrogen response gene expression on a whole-genome scale. TCGA data analysis revealed that RNF2 was positively correlated with ERα target gene expression. Further mechanistic studies showed that RNF2 was mainly localized in the nucleus and associated with ERα. The association increased ERα stability by inhibiting ERα K48-linked polyubiquitination. In conclusion, our study implicates nongenomic regulation by RNF2 on ERα protein stability and suggests that targeting RNF2 could be a promising strategy for breast cancer treatments.

Deep learning approach for cancer subtype classification using high-dimensional gene expression data

MotivationStudies have shown that classifying cancer subtypes can provide valuable information for a range of cancer research, from aetiology and tumour biology to prognosis and personalized treatment. Current methods usually adopt gene expression data to perform cancer subtype classification. However, cancer samples are scarce, and the high-dimensional features of their gene expression data are too sparse to allow most methods to achieve desirable classification results.ResultsIn this paper, we propose a deep learning approach by combining a convolutional neural network (CNN) and bidirectional gated recurrent unit (BiGRU): our approach, DCGN, aims to achieve nonlinear dimensionality reduction and learn features to eliminate irrelevant factors in gene expression data. Specifically, DCGN first uses the synthetic minority oversampling technique algorithm to equalize data. The CNN can handle high-dimensional data without stress and extract important local features, and the BiGRU can analyse deep features and retain their important information; the DCGN captures key features by combining both neural networks to overcome the challenges of small sample sizes and sparse, high-dimensional features. In the experiments, we compared the DCGN to seven other cancer subtype classification methods using breast and bladder cancer gene expression datasets. The experimental results show that the DCGN performs better than the other seven methods and can provide more satisfactory classification results.