Integrating RNA-seq and scRNA-seq to explore the biological significance of NAD + metabolism-related genes in the initial diagnosis and relapse of childhood B-cell acute lymphoblastic leukemia

Machine learning in paediatric haematological malignancies: a systematic review of prognosis, toxicity and treatment response models

BACKGROUND

Machine Learning (ML) has demonstrated potential in enhancing care in adult oncology. However, its application in paediatric haematological malignancies is still emerging, necessitating a comprehensive review of its capabilities and limitations in this area.

METHODS

A literature search was conducted through Ovid. Studies included focused on ML models in paediatric patients with haematological malignancies. Studies were categorised into thematic groups for analysis.

RESULTS

Twenty studies, primarily on leukaemia, were included in this review. Studies were organised into thematic categories such as prognoses, treatment responses and toxicity predictions. Prognostic studies showed AUC scores between 0.685 and 0.929, indicating moderate-high predictive accuracy. Treatment response studies demonstrated AUC scores between 0.840 and 0.875, reflecting moderate accuracy. Toxicity prediction studies reported high accuracy with AUC scores from 0.870 to 0.927. Only five studies (25%) performed external validation. Significant heterogeneity was noted in ML tasks, reporting formats, and effect measures across studies, highlighting a lack of standardised reporting and challenges in data comparability.

CONCLUSION

The clinical applicability of these ML models remains limited by the lack of external validation and methodological heterogeneity. Addressing these challenges through standardised reporting and rigorous external validation is needed to translate ML from a promising research tool into a reliable clinical practice component.

IMPACT

Key message: Machine Learning (ML) significantly enhances predictive models in paediatric haematological cancers, offering new avenues for personalised treatment strategies. Future research should focus on developing ML models that can integrate with real-time clinical workflows. Addition to literature: Provides a comprehensive overview of current ML applications and trends. It identifies limitations to its applicability, including the limited diversity in datasets, which may affect the generalisability of ML models across different populations.

IMPACT

Encourages standardisation and external validation in ML studies, aiming to improve patient outcomes through precision medicine in paediatric haematological oncology.

Identification and validation of NAD+ metabolism-related biomarkers in patients with diabetic peripheral neuropathy

Background

The mechanism of Nicotinamide Adenine Dinucleotide (NAD+) metabolism-related genes (NMRGs) in diabetic peripheral neuropathy (DPN) is unclear. This study aimed to find new NMRGs biomarkers in DPN.

Methods

DPN related datasets GSE95849 and GSE185011 were acquired from the Gene Expression Omnibus (GEO) database. 51 NMRGs were collected from a previous article. To explore NMRGs expression in DPN and control samples, differential expression analysis was completed in GSE95849 to obtain differentially expressed genes (DEGs), and the intersection of DEGs and NMRGs was regarded as DE-NMRGs. Next, a protein-protein interaction (PPI) network based on DE-NMRGs was constructed and biomarkers were screened by eight algorithms. Additionally, Gene Set Enrichment Analysis (GSEA) enrichment analysis was completed, biomarker-based column line graphs were constructed, lncRNA-miRNA-mRNA and competing endogenouse (ce) RNA networks were constructed, and drug prediction was completed. Finally, biomarkers expression validation was completed in GSE95849 and GSE185011.

Results

5217 DEGs were obtained from GSE95849 and 21 overlapping genes of DEGs and NMRGs were DE-NMRGs. Functional enrichment analysis revealed that DE-NMRGs were associated with glycosyl compound metabolic process. The PPI network contained 93 protein-interaction pairs and 21 nodes, with strong interactions between NMNAT1 and NAMPT, NADK and NMNAT3, ENPP3 and NUDT12 as biomarkers based on 8 algorithms. Expression validation suggested that ENPP3 and NUDT12 were upregulated in DPN samples (P < 0.05). Moreover, an alignment diagram with good diagnostic efficacy based on ENPP3 and NUDT12 were identified was constructed. GSEA suggested that ENPP3 was enriched in Toll like receptor (TLR) pathway, NUDT12 was enriched in maturity onset diabetes of the young and insulin pathway. Furthermore, 18 potential miRNAs and 36 Transcription factors (TFs) were predicted and the miRNA-mRNA-TF networks were constructed, suggesting that ENPP3 might regulate hsa-miR-34a-5p by affecting MYNN. The ceRNA network suggested that XLOC_013024 might regulate hsa-let-7b-5p by affecting NUDT12. 15 drugs were predicted, with 8 drugs affecting NUDT12 such as resveratrol, and 13 drugs affecting ENPP3 such as troglitazone.

Conclusion

ENPP3 and NUDT12 might play key roles in DPN, which provides reference for further research on DPN.

Integrating scRNA-seq to explore novel macrophage infiltration-associated biomarkers for diagnosis of heart failure

OBJECTIVE

Inflammation and immune cells are closely intertwined mechanisms that contribute to the progression of heart failure (HF). Nonetheless, there is a paucity of information regarding the distinct features of dysregulated immune cells and efficient diagnostic biomarkers linked with HF. This study aims to explore diagnostic biomarkers related to immune cells in HF to gain new insights into the underlying molecular mechanisms of HF and to provide novel perspectives for the detection and treatment of HF.

METHOD

The CIBERSORT method was employed to quantify 22 types of immune cells in HF and normal subjects from publicly available GEO databases (GSE3586, GSE42955, GSE57338, and GSE79962). Machine learning methods were utilized to screen for important cell types. Single-cell RNA sequencing (GSE145154) was further utilized to identify important cell types and hub genes. WGCNA was employed to screen for immune cell-related genes and ultimately diagnostic models were constructed and evaluated. To validate these predictive results, blood samples were collected from 40 normal controls and 40 HF patients for RT-qPCR analysis. Lastly, key cell clusters were divided into high and low biomarker expression groups to identify transcription factors that may affect biomarkers.

RESULTS

The study found a noticeable difference in immune environment between HF and normal subjects. Macrophages were identified as key immune cells by machine learning. Single-cell analysis further showed that macrophages differed dramatically between HF and normal subjects. This study revealed the existence of five subsets of macrophages that have different differentiation states. Based on module genes most relevant to macrophages, macrophage differentiation-related genes (MDRGs), and DEGs in HF and normal subjects from GEO datasets, four genes (CD163, RNASE2, LYVE1, and VSIG4) were identified as valid diagnostic markers for HF. Ultimately, a diagnostic model containing two hub genes was constructed and then validated with a validation dataset and clinical samples. In addition, key transcription factors driving or maintaining the biomarkers expression programs were identified.

CONCLUSION

The analytical results and diagnostic model of this study can assist clinicians in identifying high-risk individuals, thereby aiding in guiding treatment decisions for patients with HF.

Integrative Analysis of Single-Cell and Bulk RNA Sequencing Reveals Prognostic Characteristics of Macrophage Polarization-Related Genes in Lung Adenocarcinoma

Background

Lung adenocarcinoma (LUAD) is a group of cancers with poor prognosis. The combination of single-cell RNA sequencing (scRNA-seq) and bulk RNA sequencing (RNA-seq) can identify important genes involved in cancer development and progression from a broader perspective.

Methods

The scRNA-seq data and bulk RNA-seq data of LUAD were downloaded from the Gene Expression Omnibus (GEO) database and the Cancer Genome Atlas (TCGA) database. Analyzing scRNA-seq for core cells in the GSE131907 dataset, and the uniform manifold approximation and projection (UMAP) was used for dimensionality reduction and cluster identification. Macrophage polarization-associated subtypes were acquired from the TCGA-LUAD dataset after analysis, followed by further identification of differentially expressed genes (DEGs) in the TCGA-LUAD dataset (normal/LUAD tissue samples, two subtypes). Venn diagrams were utilized to visualize differentially expressed and highly variable macrophage polarization-related genes. Subsequently, a prognostic risk model for LUAD patients was constructed by univariate Cox and Least Absolute Shrinkage and Selection Operator (LASSO), and the model was investigated for stability in the external data GSE72094. After analyzing the correlation between the trait genes and significantly mutated genes, the immune infiltration between the high/low-risk groups was then examined. The Monocle package was applied to analyze the pseudo-temporal trajectory analysis of different cell clusters in macrophage clusters. Subsequently, cell clusters of data macrophages were selected as key cell clusters to explore the role of characteristic genes in different cell populations and to identify transcription factors (TFs) that affect signature genes. Finally, qPCR were employed to validate the expression levels of prognosis signature genes in LUAD.

Results

424 macrophage highly variable genes, 3920 DEGs, and 9561 DEGs were obtained from macrophage clusters, the macrophage polarization-related subtypes, and normal/LUAD tissue samples, respectively. Twenty-eight differentially expressed and highly mutated MPRGs were obtained. A prognostic risk model with 7 DE-MPRGs (RGS13, ADRB2, DDIT4, MS4A2, ALDH2, CTSH, and PKM) was constructed. This prognostic model still has a good prediction effect in the GSE72094 dataset. ZNF536 and DNAH9 were mutated in the low-risk group, while COL11A1 was mutated in the high-risk group, and they were highly correlated with the characteristic genes. A total of 11 immune cells were significantly different in the high/low-risk groups. Five cell types were again identified in the macrophage cluster, and then NK cells: CD56hiCD62L+ differentiated earlier and were present mainly on 2 branches. While macrophages were present on 2 branches and differentiated later. It was found that the expression levels of BCLAF1 and MAX were higher in cluster 1, which might be the TFs affecting the expression of the characteristic genes. Moreover, qPCR confirmed that the expression of the prognosis genes was generally consistent with the results of the bioinformatic analysis.

Conclusion

Seven MPRGs (RGS13, ADRB2, DDIT4, MS4A2, ALDH2, CTSH, and PKM) were identified as prognostic genes for LUAD and revealed the mechanisms of MPRGs at the single-cell level.

Identification of ULK1 as a novel mitophagy-related gene in diabetic nephropathy

BACKGROUND

Accumulating evidence indicates that mitophagy is crucial for the development of diabetic nephropathy (DN). However, little is known about the key genes involved. The present study is to identify the potential mitophagy-related genes (MRGs) in DN.

METHODS

Five datasets were obtained from the Gene Expression Omnibus (GEO) database and were split into the training and validation set. Then the differentially expressed MRGs were screened and further analyzed for GO and KEGG enrichment. Next, three algorithms (SVM-RFE, LASSO and RF) were used to identify hub genes. The ROC curves were plotted based on the hub genes. We then used the CIBERSORT algorithm to assess the infiltration of 22 types of immune cells and explore the correlation between hub genes and immune cells. Finally, the Nephroseq V5 tool was used to analyze the correlation between hub genes and GFR in DN patients.

RESULTS

Compared with the tubulointerstitium, the expression of MRGs was more noticeably varied in the glomeruli. Twelve DE-MRGs were identified in glomerular samples, of which 11 genes were down-regulated and only MFN1 was up-regulated. GO and KEGG analysis indicated that several enrichment terms were associated with changes in autophagy. Three genes (MFN1, ULK1 and PARK2) were finally determined as potential hub genes by three algorithms. In the training set, the AUROC of MFN1, ULK1 and PARK2 were 0.839, 0.906 and 0.842. However, the results of the validation set demonstrated that MFN1 and PARK2 had no significant difference in distinguishing DN samples from healthy controls, while the AUROC of ULK1 was 0.894. Immune infiltration analysis using CIBERSORT showed that ULK1 was positively related to neutrophils, whereas negatively related to M1 and M2 macrophages. Finally, ULK1 was positively correlated with GFR in Nephroseq database.

CONCLUSIONS

ULK1 is a potential biomarker for DN and may influence the development of diabetic nephropathy by regulating mitophagy.