Machine learning and BP neural network revealed abnormal B cell infiltration predicts the survival of lung cancer patients

Tu, Pinghua; Li, Xinjun; Cao, Lingli; Zhong, Ming; Xie, Zhibin; Wu, Zuoqiang

doi:10.3389/fonc.2022.882018

Cited by 4 publications

(2 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using ssGSEA by interrogating the expression levels of reported specific genes of adaptive and innate immune cell types, we computed the infiltration levels of 28 immune cell types per sample. Next, we integrated Lasso and RF algorithms to obtain key immune cell types from 28 immune cell types [50]. Lasso and RF were constructed under 10-fold cross-validation, and the setting parameters of RF were 800 decision trees.…”

Section: Evaluation Of Immune Cell Characteristics In Hf Via Ssgsea A...mentioning

confidence: 99%

Identifying OGN as a Biomarker Covering Multiple Pathogenic Pathways for Diagnosing Heart Failure: From Machine Learning to Mechanism Interpretation

Zhu,

Chen,

2024

Biomolecules

View full text Add to dashboard Cite

Background: The pathophysiologic heterogeneity of heart failure (HF) necessitates a more detailed identification of diagnostic biomarkers that can reflect its diverse pathogenic pathways. Methods: We conducted weighted gene and multiscale embedded gene co-expression network analysis on differentially expressed genes obtained from HF and non-HF specimens. We employed a machine learning integration framework and protein–protein interaction network to identify diagnostic biomarkers. Additionally, we integrated gene set variation analysis, gene set enrichment analysis (GSEA), and transcription factor (TF)-target analysis to unravel the biomarker-dominant pathways. Leveraging single-sample GSEA and molecular docking, we predicted immune cells and therapeutic drugs related to biomarkers. Quantitative polymerase chain reaction validated the expressions of biomarkers in the plasma of HF patients. A two-sample Mendelian randomization analysis was implemented to investigate the causal impact of biomarkers on HF. Results: We first identified COL14A1, OGN, MFAP4, and SFRP4 as candidate biomarkers with robust diagnostic performance. We revealed that regulating biomarkers in HF pathogenesis involves TFs (BNC2, MEOX2) and pathways (cell adhesion molecules, chemokine signaling pathway, cytokine–cytokine receptor interaction, oxidative phosphorylation). Moreover, we observed the elevated infiltration of effector memory CD4+ T cells in HF, which was highly related to biomarkers and could impact immune pathways. Captopril, aldosterone antagonist, cyclopenthiazide, estradiol, tolazoline, and genistein were predicted as therapeutic drugs alleviating HF via interactions with biomarkers. In vitro study confirmed the up-regulation of OGN as a plasma biomarker of HF. Mendelian randomization analysis suggested that genetic predisposition toward higher plasma OGN promoted the risk of HF. Conclusions: We propose OGN as a diagnostic biomarker for HF, which may advance our understanding of the diagnosis and pathogenesis of HF.

show abstract

Section: Evaluation Of Immune Cell Characteristics In Hf Via Ssgsea A...mentioning

confidence: 99%

Identifying OGN as a Biomarker Covering Multiple Pathogenic Pathways for Diagnosing Heart Failure: From Machine Learning to Mechanism Interpretation

Zhu,

Chen,

2024

Biomolecules

View full text Add to dashboard Cite

show abstract

“…The investigation showed that FAM83A presents prognostic and predictive properties supported by machine-learning algorithms. Therefore, the unusual tumor permeation of naive B cells monitored by FAM83A may reveal robust biomarkers that preview the clinical behavior of patients with non-small cell lung cancer (10).…”

mentioning

confidence: 99%

Editorial: Multiplexed image analysis for translational research project applications

Capelozzi

Parra

2022

Front. Oncol.

View full text Add to dashboard Cite

Editorial on the Research Topic Multiplexed image analysis for translational research project applicationsWith the continuous improvement in our understanding of immuno-oncology and in the therapeutic arsenal for establishing personalized treatment for patients, tumor biomarkers targeting specific disease patterns are necessary. The evaluation of protein expression employing immunostaining of formalin-fixed, paraffin-embedded (FFPE) specimens is an ancillary technique for guiding clinical decisions in multiple situations, such as tumor stratification, evaluation of disease recurrence, precise mutation discovery, and subsequent targeting of treatment towards specific biomarkers. However, standard enzyme-based chromogenic immunohistochemistry (IHC) is a classic technology to study tumor microenvironments has limitations such as the use of separate slides per marker or challenges to analyze markers that are expressed by different cell populations with digital image analysis (1). In contrast to standard bright-field IHC staining, multiplexed image analysis technologies offer the opportunity to observe several markers independently in one tissue section without staining each marker on separate slides (2). Multiplexed analysis also allows accurate protein co-localization, stain adjustment, objective scoring, and identification of cut-off values for each biomarker. These properties are critical in low immune-stained areas, which represent difficult areas to evaluate on immunostained glass slides and may be misinterpreted as unfavorable because of camouflaging by the hematoxylin counterstain (3). In this Research Topic, we highlight a series of publications focusing on the applicability of the commercial modalities in most common use multiplexed imaging modalities to enable prospective users to conduct translational research and tumor immune environment characterization.Several highly multiplexed tissue imaging technologies have also emerged involving cyclic immunofluorescence (IF), tyramide-based IHC/IF, epitope-targeted mass spectrometry, or RNA detection, allowing detailed studies of cell structure, functionalFrontiers in Oncology frontiersin.org 01

show abstract

Sense and anti-sense: Role of FAM83A and FAM83A-AS1 in Wnt, EGFR, PI3K, EMT pathways and tumor progression

Zhao,

Li,

Zhang

et al. 2024

Biomedicine & Pharmacotherapy

View full text Add to dashboard Cite

Machine learning and BP neural network revealed abnormal B cell infiltration predicts the survival of lung cancer patients

Cited by 4 publications

References 37 publications

Identifying OGN as a Biomarker Covering Multiple Pathogenic Pathways for Diagnosing Heart Failure: From Machine Learning to Mechanism Interpretation

Identifying OGN as a Biomarker Covering Multiple Pathogenic Pathways for Diagnosing Heart Failure: From Machine Learning to Mechanism Interpretation

Editorial: Multiplexed image analysis for translational research project applications

Sense and anti-sense: Role of FAM83A and FAM83A-AS1 in Wnt, EGFR, PI3K, EMT pathways and tumor progression

Contact Info

Product

Resources

About