DECO: decompose heterogeneous population cohorts for patient stratification and discovery of sample biomarkers using omic data profiling

Campos‐Laborie, Francisco J.; Risueño, Alberto; Ortiz-Estévez, María; Rosón-Burgo, Beatriz; Droste, Conrad; Fontanillo, Celia; Loos, Remco; Sánchez‐Santos, José Manuel; Trotter, Matthew; Rivas, Javier De Las

doi:10.1093/bioinformatics/btz148

Cited by 10 publications

(18 citation statements)

References 72 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To bolster the validation of PHet’s discriminative capabilities, we conducted a comprehensive evaluation of 25 algorithms (outlined in Section 3 and Table 1) using two sets of simulated data, specifically tailored to test DE capability of each algorithm. These datasets correspond to the ‘minority’ and ‘mixed’ model schemes, as proposed by Campos and colleagues 43 , and are designed to capture sample heterogeneity under the supervised settings. In the ‘minority’ model, a small fraction of case samples exhibited changes in specific features, while the ‘mixed’ model displayed substantial intra-group variation in both case and control samples for those features.…”

Section: Resultsmentioning

confidence: 99%

“…Several approaches, such as DIDS 42 and DECO 43 , have been developed to extract relevant features from expression data and to identify subtypes. However, these methods have substantial limitations in capturing subtype-related features because they focus on a specific type of outlier features, which means they may miss important information that could be useful for subtyping.…”

Section: Discussionmentioning

confidence: 99%

“…The performance of the PHet algorithm was evaluated in comparison with four DE feature analysis tools and their variants: the standard Student t-test 62 , Wilcoxon rank-sum test 63 , Kolmogorov–Smirnov test 64 , and LIMMA 65 ; a dispersion-based method and its variants 45 ; an IQR-based approach and its variations 47 ; and seven outlier detection algorithms: COPA 66 , OS 38 , ORT 39 , MOST 40 , LSOSS 41 , DIDS 42 , and DECO 43 . In addition, we also assessed the performance of PHet using the ΔDispersion metric instead of ΔIQR.…”

Section: Methodsmentioning

confidence: 99%

“…This method was designed to identify subtypes within cancer patients by focusing on the genes whose expression profiles exhibit minor outliers. This pioneering work inspired a series of studies that improved COPA with more sophisticated statistical and machine learning techniques for subgroup discovery within each experimental condition [38][39][40][41][42][43] . These methods involve ranking features based on statistical metrics that gauge the degree of abnormal expression across two experimental scenarios.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Heterogeneity-Preserving Discriminative Feature Selection for Subtype Discovery

Basher

Hallinan

Lee

2023

Preprint

View full text Add to dashboard Cite

Subtype discovery is crucial to disease diagnosis and targeted therapy as each cell or patient can display a wide range of responses to specific treatments. It is also vital to investigate the heterogeneity, or diverse variety, of disease states to better understand pathological processes. While there are many new opportunities for subtype discovery due to the explosion of single-cell data (single-cell RNA-seq, proteomic, and imaging datasets), selecting features for disease subtyping from high-dimensional datasets is challenging. Feature selection is the process of reducing the number of features for downstream computational analyses. Most feature selection algorithms are focused on identifying features to aid in the classification of known disease phenotypes. However, this effectively eliminates heterogeneous features and collapses the disease feature space, hindering valuable subtyping. Our work aimed to identify feature sets that preserve heterogeneity while maintaining the discrimination of known disease states. We initially applied a data-driven approach to determine the statistical characteristics of features essential in preserving heterogeneity by combining feature clustering and deep metric learning. This analysis revealed that features with a significant difference in interquartile range (IQR) between classes could contain critical subtype information. Utilizing this knowledge, we developed a statistical method, PHet (Preserving Heterogeneity), that performs recurrent sub-sampling differential analysis of IQR between classes to identify a minimal set of heterogeneity-preserving features while maximizing the quality of subtype clustering. Using public datasets of microarray and single-cell RNA-seq, we demonstrated that PHet effectively identifies disease subtypes and outperforms the previous outlier-based methods. In summary, our study provides a novel feature selection method for disease subtyping, which will enable not only personalized medicine but also a detailed understanding of the underlying heterogeneous mechanisms of diseases.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Heterogeneity-Preserving Discriminative Feature Selection for Subtype Discovery

Basher

Hallinan

Lee

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The application of omic technologies in cancer research lead to the generation of thousands of data that can only be useful if advanced computational and statistical methods are available for their analysis. De Las Rivas' group developed this type of methods, including the DECO method (decomposing heterogeneous cohorts using omics data profiling) [33]. This new method allows to find significant association between biological features (biomarkers) and samples in large scale omics data.…”

Section: Session III Functional Genomics In Clinical Practicementioning

confidence: 99%

From single gene analysis to single cell profiling: a new era for precision medicine

Martino

Meschini²,

Scotlandi

et al. 2020

J Exp Clin Cancer Res

View full text Add to dashboard Cite

Molecular profiling of DNA and RNA has provided valuable new insights into the genetic basis of non-malignant and malignant disorders, as well as an increased understanding of basic mechanisms that regulate human disease. Recent technological advances have enabled the analyses of alterations in gene-based structure or function in a comprehensive, high-throughput fashion showing that each tumor type typically exhibits distinct constellations of genetic alterations targeting one or more key cellular pathways that regulate cell growth and proliferation, evasion of the immune system, and other aspects of cancer behavior. These advances have important implications for future research and clinical practice in areas as molecular diagnostics, the implementation of gene or pathwaydirected targeted therapy, and the use of such information to drive drug discovery. The 1st international and 32nd Annual Conference of Italian Association of Cell Cultures (AICC) conference wanted to offer the opportunity to match technological solutions and clinical needs in the era of precision medicine.

show abstract