Abstract Motivation Recent breakthroughs of single-cell RNA sequencing (scRNA-seq) technologies offer an exciting opportunity to identify heterogeneous cell types in complex tissues. However, the unavoidable biological noise and technical artifacts in scRNA-seq data as well as the high dimensionality of expression vectors make the problem highly challenging. Consequently, although numerous tools have been developed, their accuracy remains to be improved. Results Here, we introduce a novel clustering algorithm and tool RCSL (Rank Constrained Similarity Learning) to accurately identify various cell types using scRNA-seq data from a complex tissue. RCSL considers both local similarity and global similarity among the cells to discern the subtle differences among cells of the same type as well as larger differences among cells of different types. RCSL uses Spearman’s rank correlations of a cell’s expression vector with those of other cells to measure its global similarity, and adaptively learns neighbour representation of a cell as its local similarity. The overall similarity of a cell to other cells is a linear combination of its global similarity and local similarity. RCSL automatically estimates the number of cell types defined in the similarity matrix, and identifies them by constructing a block-diagonal matrix, such that its distance to the similarity matrix is minimized. Each block-diagonal submatrix is a cell cluster/type, corresponding to a connected component in the cognate similarity graph. When tested on 16 benchmark scRNA-seq datasets in which the cell types are well-annotated, RCSL substantially outperformed six state-of-the-art methods in accuracy and robustness as measured by three metrics. Availability The RCSL algorithm is implemented in R and can be freely downloaded at https://github.com/QinglinMei/RCSL . Contact guojunsdu@gmail.com , zcsu@uncc.edu Supplementary information Supplementary data are available at Bioinformatics online.
Biclustering has emerged as a promising approach for analyzing high-dimensional expression data, offering unique advantages in uncovering localized co-expression patterns that traditional clustering methods often miss and thus facilitating advancements in complex disease research and other biomedical applications. However, state-of-the-art methods identify distinct patterns at the expense of losing information about specific patterns, some of which have been used to define cancer subtypes or reflect the progression of a disease or cellular processes. Additionally, these methods exhibit poor effectiveness in noisy environments. To address these limitations, we propose the bucket trend-preserving (BTP) pattern, a novel generalization of existing patterns. And we have developed an algorithm, TransBic, to extract significant biclusters of BTP-patterns. Specifically, TransBic transforms the problem into identifying common multipartite acyclic tournament subdigraphs shared by distinct subsets of acyclic tournament digraphs derived from a given expression matrix. Compared with prominent tools, TransBic demonstrates superior performance in identifying biclusters of all non-row-constant patterns, especially under noise and data fluctuations. Furthermore, TransBic successfully identifies the most disease-related pathways for type 2 diabetes (T2D), colorectal cancer, hepatocellular carcinoma, and breast cancer, outperforming other tools in this regard. Different from previous generalizations, BTP-patterns capture specific up-regulation and down-regulation dynamics. Through targeted analysis of BTP-patterns in T2D expression data, TransBic uncovers biological processes affected by disease risk factors, extending the application of trend-preserving biclustering in expression data analysis.
Recent breakthroughs of single-cell RNA sequencing (scRNA-seq) technologies offer an exciting opportunity to identify heterogeneous cell types in complex tissues. However, the unavoidable biological noise and technical artifacts in scRNA-seq data as well as the high dimensionality of expression vectors make the problem highly challenging. Consequently, although numerous tools have been developed, their accuracy remains to be improved.Here, we introduce a novel clustering algorithm and tool RCSL (Rank Constrained Similarity Learning) to accurately identify various cell types using scRNA-seq data from a complex tissue. RCSL considers both local similarity and global similarity among the cells to discern the subtle differences among cells of the same type as well as larger differences among cells of different types. RCSL uses Spearman's rank correlations of a cell's expression vector with those of other cells to measure its global similarity, and adaptively learns neighbor representation of a cell as its local similarity. The overall similarity of a cell to other cells is a linear combination of its global similarity and local similarity. RCSL automatically estimates the number of cell types defined in the similarity matrix, and identifies them by constructing a block-diagonal matrix, such that its distance to the similarity matrix is minimized. Each block-diagonal submatrix is a cell cluster/type, corresponding to a connected component in the cognate similarity graph. When tested on 16 benchmark scRNA-seq datasets in which the cell types are well-annotated, RCSL substantially outperformed six state-of-the-art methods in accuracy and robustness as measured by three metrics.The RCSL algorithm is implemented in R and can be freely downloaded at https://cran.r-project.org/web/packages/RCSL/index.html.Supplementary data are available at Bioinformatics online.
Single-cell RNA sequencing (scRNA-seq) technologies have been driving the development of algorithms of clustering heterogeneous cells. We introduce a novel clustering algorithm scQA, which can effectively and efficiently recognize different cell types via qualitative and quantitative analysis. It iteratively extracts quasi-trend-preserved genes to conform a consensus by representing expression patterns with dropouts qualitatively and quantitatively, and, then automatically clusters cells using a new label propagation strategy without specifying the number of cell types in advance. Validated on 20 public scRNA-seq datasets, scQA consistently outperformed 9 salient tools in both accuracy and efficiency across 16 out of 20 datasets tested, and ranked top 2 or 3 across the other 4 datasets. Furthermore, we demonstrate scQA can extract informative genes in both perspectives of biology and data wise by performing consensus, allowing genes used for landmark construction multiple characteristics, which is essential for clustering cells accurately. Overall, scQA could be a useful tool for discovery of cell types that can be integrated into general scRNA-seq analyses.
Abstract Background Cancer is widely regarded as a complex disease primarily driven by genetic mutations. A critical concern and significant obstacle lies in discerning driver genes amid an extensive array of passenger genes. Findings We present a new method termed DriverMP for effectively prioritizing altered genes on a cancer-type level by considering mutated gene pairs. It is designed to first apply nonsilent somatic mutation data, protein‒protein interaction network data, and differential gene expression data to prioritize mutated gene pairs, and then individual mutated genes are prioritized based on prioritized mutated gene pairs. Application of this method in 10 cancer datasets from The Cancer Genome Atlas demonstrated its great improvements over all the compared state-of-the-art methods in identifying known driver genes. Then, a comprehensive analysis demonstrated the reliability of the novel driver genes that are strongly supported by clinical experiments, disease enrichment, or biological pathway analysis. Conclusions The new method, DriverMP, which is able to identify driver genes by effectively integrating the advantages of multiple kinds of cancer data, is available at https://github.com/LiuYangyangSDU/DriverMP. In addition, we have developed a novel driver gene database for 10 cancer types and an online service that can be freely accessed without registration for users. The DriverMP method, the database of novel drivers, and the user-friendly online server are expected to contribute to new diagnostic and therapeutic opportunities for cancers.
Background: Tumor classification is one of the most important applications of gene expression data. Due to high dimensionality in microarray data, dimensionality reduction plays a crucial role in tumor classification based on gene expression profiles. Objective: The primary objective of this study is to increase the accuracy of tumor classification by reducing the dimensionality of gene expression data with feature extraction methods. Method: In this paper, we propose a novel supervised feature extraction method for tumor classification called discriminant hybrid structure preserving projections. The proposed method utilizes hybrid representation to efficiently characterize the structure of gene expression data, where both neighbor representation and sparse representation are taken into account. Specifically, our algorithm enhances the data separability after dimensionality reduction by simultaneously minimizing the within-class distance and maximizing the between-class distance. Moreover, it employs an imbalanced adjustment factor during the extraction process to overcome the class imbalance problem in tumor datasets. Results: Experiments on five publicly available tumor datasets demonstrate the effectiveness of the proposed method in comparison with a number of state-of-the-art feature extraction and feature selection methods. Conclusion: The proposed algorithm can enhance the separability of data after projections and thus improve the tumor classification accuracy of gene expression data. Keywords: Tumor classification, gene expression data, dimensionality reduction, feature extraction, neighbor representation, sparse representation.
Single-cell RNA sequencing technologies have been pivotal in advancing the development of algorithms for clustering heterogeneous cell populations. Existing methods for utilizing scRNA-seq data to identify cell types tend to neglect the beneficial impact of dropout events and perform clustering focusing solely on quantitative perspective. Here, we introduce a novel method named scQA, notable for its ability to concurrently identify cell types and cell type-specific key genes from both qualitative and quantitative perspectives. In contrast to other methods, scQA not only identifies cell types but also extracts key genes associated with these cell types, enabling bidirectional clustering for scRNA-seq data. Through an iterative process, our approach aims to minimize the number of landmarks to approximately a dozen while maximizing the inclusion of quasi-trend-preserved genes with dropouts both qualitatively and quantitatively. It then clusters cells by employing an ingenious label propagation strategy, obviating the requirement for a predetermined number of cell types. Validated on 20 publicly available scRNA-seq datasets, scQA consistently outperforms other salient tools. Furthermore, we confirm the effectiveness and potential biological significance of the identified key genes through both external and internal validation. In conclusion, scQA emerges as a valuable tool for investigating cell heterogeneity due to its distinctive fusion of qualitative and quantitative facets, along with bidirectional clustering capabilities. Furthermore, it can be seamlessly integrated into border scRNA-seq analyses. The source codes are publicly available at https://github.com/LD-Lyndee/scQA.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
