Skip to main content
Download PDF
Download ePub

Machine learning-based integration identifies the ferroptosis hub genes in nonalcoholic steatohepatitis

Lipids in Health and Disease volume 23, Article number: 23 (2024) Cite this article

Abstract

Background

Ferroptosis, is characterized by lipid peroxidation of fatty acids in the presence of iron ions, which leads to cell apoptosis. This leads to the disruption of metabolic pathways, ultimately resulting in liver dysfunction. Although ferroptosis is linked to nonalcoholic steatohepatitis (NASH), understanding the key ferroptosis-related genes (FRGs) involved in NASH remains incomplete. NASH may be targeted therapeutically by identifying the genes responsible for ferroptosis.

Methods

To identify ferroptosis-related genes and develop a ferroptosis-related signature (FeRS), 113 machine-learning algorithm combinations were used.

Results

The FeRS constructed using the Generalized Linear Model Boosting algorithm and Gradient Boosting Machine algorithms exhibited the best prediction performance for NASH. Eight FRGs, with ZFP36 identified by the algorithms as the most crucial, were incorporated into in FeRS. ZFP36 is significantly enriched in various immune cell types and exhibits significant positive correlations with most immune signatures.

Conclusion

ZFP36 is a key FRG involved in NASH pathogenesis.

Introduction

There is mounting evidence implicating non-alcoholic fatty liver disease (NAFLD) result in persistent liver damage [1, 2]. The increasing incidence of NAFLD affects human survival time [3]. In the early stages, NAFLD manifests as nonalcoholic fatty liver disease [4]; nevertheless, advanced-stage NAFLD is characterized by necroinflammatory responses and increased hepatic fat accumulation. NASH can result from the accumulation of hepatic fat and the infiltration of inflammatory cytokines [5]. NASH liver cells can develop fibrosis, hardening, and even cancer if no intervention is performed [6]; thus, early identification of NASH is crucial. NASH is traditionally diagnosed using a liver biopsy [7], but its invasive nature limits its clinical application. Another challenge in the management of NASH is the lack of effective treatments. Although lifestyle modifications and increased physical activity are considered primary therapeutic measures for NASH, research suggests that their outcomes are suboptimal [8]. Thus, the search for therapeutic targets in NASH remains a critical endeavor.

Ferroptosis entails the activation of reactive oxygen species (ROS) and lipid peroxidation by iron ions [9]. Ferroptosis likely contributes to the development of NASH because of the significant role of the liver in storing iron ions and regulating lipid metabolism [10]. NASH treatment may be enhanced by understanding the core ferroptosis genes in patients with NASH.

In machine learning, predictive models are constructed by identifying complex patterns in datasets or by correlating data with predictions [11]. This analytical discipline is useful for the development of optimal predictive models for disease diagnosis. Machine learning has been shown to provide robust risk models within clinical datasets, enabling the delineation of distinct patient cohorts based on clinical data [12, 13].

This study hypothesises that the innovative integration of multiple machine learning algorithms to identify FRGs would develop a non-invasive diagnostic model to better serve patients with NASH.

Methods

Obtaining GEO data

Seven cohorts containing information from NASH patients (GSE130970, GSE89632, GSE61260, GSE126848, GSE48452, GSE164760, and GSE63067) were selected from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). Gene matrices were calculated using the platform files associated with each GEO dataset. The following table presents the platform file information for the seven GEO datasets. A merged dataset was created by merging the GSE130970 and GSE89632 datasets, after removing batch effects. The remaining five datasets were considered the external validation sets. Additionally, to increase the number of external validation sets, two additional validation cohorts (all-testing set cohorts and sample cohorts) were created (Supplementary Table 1).

Filtering differentially expressed genes (DEGs) associated with Ferroptosis

The DEGs between the two groups were screened based on |logFC| > 0.585 and corrected P-values < 0.05. Additionally, 484 FRGs are available in the FerrDb V2 database (http://www.zhounan.org/ferrdb/current/) [14]. These genes were combined to identify 16 DEGs. The comprehensive bioinformatics database Metascape (https://metascape.org/gp/index.html#/main/step1) exhibits several enrichment analysis tools and resources [15]. The Metascape database was used to determine possible biological pathways and functional categories associated with ferroptosis-related DEGs.

Constructing a ferroptosis-related signature (FeRS)

Twelve machine learning algorithms were selected for this study for the binary classification variables. The Enet, Ridge, Stepglm, and Least Absolute Shrinkage and Selection Operator (LASSO) algorithms are typically applied for regression problems, particularly when there are many features, to select the most important features and reduce the risk of overfitting. The Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), glmBoost, Partial Least Squares Regression Generalized Linear Model (plsRglm), Random Forest, GBM, extreme Gradient Boosting (XGBoost), and Naive Bayes are all suitable for classification tasks in this study. A total of 113 algorithms were developed from the 12 machine learning algorithms. Two measures are implemented to mitigate the risk of overfitting. A cross-validation approach was used in the first step, with one algorithm for variable selection and the other for constructing the classification prediction model for each combination algorithm. Another approach involves increasing the number of validation datasets.

The areas under the receiver operating characteristic (ROC) curves were evaluated for the 113 machine-learning algorithms in eight cohorts. A heat map was used to show the results of the model. The best algorithm was selected from among 113 algorithms. Model genes filtered using the best algorithm were used to calculate the ROC values for diagnosing NASH in the eight cohorts.

Pathway enrichment and immune infiltration analysis

GSVA analysis identified significantly enriched biological pathways [16]. Additionally, it was used to explore enriched biological pathways when the model genes were upregulated or downregulated. Differences in the enrichment of immune features and immune cells were investigated in model genes.

GENEMANIA was applied to identify genes that interact with the model genes [17]. Friendship analysis is a network topology-based analytical method used to explore interactions and relationships between genes or proteins in biological processes [18]. It relies on semantic similarity measurements based on GO, which means that associations are analyzed using the functional information of genes. Friendship analysis can help select the most significant genes from a large set of DEGs. Therefore, it was employed to evaluate the most crucial genes among the model genes.

Single-cell analysis

The GSE159977 and GSE129516 datasets were downloaded. The former was separated into two groups, GSE159977(1) and GSE159977(2), each containing five human NASH samples. GSE129516 consists of three NASH mouse samples. To analyze the GSE159977(1), GSE159977(2), and GSE129516 datasets, the “Seurat” package was applied separately. First, the “PercentageFeatureSet()” function and all gene sets starting with MT- for mitochondrial quality control analysis were used to assess the proportion of mitochondrial genes in each cell to identify and exclude low-quality cells. To determine which features within the dataset displayed a high level of inter-cell variability, gene expression values were normalized using the “LogNormalize” method. For each dataset, 1500 features were selected for the PCA dimensionality reduction analysis, and the “JackStrawPlot()” and “ElbowPlot()” functions were used to determine the dimensionality of the dataset. Subsequently, cell clustering was performed, and the results were visualized using either the t-distributed stochastic neighborhood embedding method or the uniform surface approximation and projection method. Finally, the “SingleR” package was used to annotate different clusters as distinct immune cells. Additionally, differences in the expression of model genes among different immune cells were determined.

Regulation factors of hub genes

The NetworkAnalyst platform was used in this investigation (https://www.networkanalyst.ca/) [19]. The JASPAR database was used to elucidate the regulatory network; the JASPAR database was selected and examined [20]. Furthermore, miRTarBase v8.0, was used to investigate miRNAs associated with the model genes [21]. Drug-targeting model genes were analyzed using the Drug-Gene Interaction database (DGIdb; http://dgidb.org/) [22]. The data were visualized using the Cytoscape software suite [23].

Quantitative real-time polymerase chain reaction (qPCR)

Eleven pairs of normal and NASH liver tissues (NASH score > 5) were collected from the hospital, and qPCR was performed. The RNA was extracted from the samples and then reverse-transcribed into cDNA. Relative expression of the model genes was standardized with GAPDH and evaluated using the 2-ΔΔCt method. Box plots were generated with the GraphPad Prism software (version 9.0.0). Primers and reagents used are showed in Supplementary Table 2.

Statistical analysis

The Mann-Whitney U-test was applied to compare the two subgroups that did not satisfy the t-test. Statistical significance was set at P < 0.05. 3. The concept of the full text is presented in Fig. 1.

Fig. 1
figure 1

The flow chart of the research

Full size image

Results

Cohort integration

Healthy and NASH samples from seven datasets were analyzed (Fig. 2A,B). After removing the batch effect, GSE130970 and GSE89632 were combined to create a training set (Fig. 2C-F).

Fig. 2
figure 2

The process of data integration. A-B Statistics on the number of samples. C, D Before and after sample integration. E-F Principal component analysis (PCA) plots before and after the removal of the batch effect

Full size image

Sixteen ferroptosis-related DEGs associated with metabolism

Four hundred seventy five DEGs were observed. A total of 271 genes were highly expressed in NASH, whereas the opposite was true for 204 genes (Fig. 3A, B). Sixteen ferroptosis-related DEGs were identified when these 475 DEGs intersected with 484 FRGs (Fig. 3C). Notably, in the NASH group, six genes were significantly upregulated, while 10 genes were significantly opposite (Fig. 3D).

Fig. 3
figure 3

It was determined that 16 genes were involved. (A) Across the two groups, 475 differential genes (DEGs) were found. (B) Expression landscape of DEGs. (C) Sixteen common genes were recognized. (D) Expression patterns of the genes. (E, F) Biological pathway enrichment analysis. (G) Disease enrichment analysis

Full size image

These 16 genes were significantly enriched in pathways associated with “alpha-linolenic acid metabolism,” “adipocyte differentiation,” “regulation of adipocyte differentiation,” and “circadian rhythm” (Fig. 3E, F). Consequently, the metabolic pathways are intertwined with NASH development in a complex manner. There was also a significant association between these genes and the etiologies of NASH, such as “fatty liver disease,” “lipid metabolism abnormalities,” and “insulin resistance” (Fig. 3G).

Development of an FeRS

After the calculations, “glmBoost+GBM” yielded the best predictive performance among the 113 algorithms tested. It was constructed by applying the “glmBoost+GBM” algorithm to an FeRS consisting of eight FRGs (GRIA3, NR4A1, FABP4, IL6, FADS2, ZFP36, DHODH, and PRKAA2). The ROC values for diagnosing NASH using FeRS in eight sets (Merge cohort, GSE61260 cohort, GSE126848 cohort, GSE63067 cohort, GSE48452 cohort, GSE164760 cohort, All Testing Sets cohort, and All Samples cohort) were 1.000, 0.817, 0.906, 0.889, 0.881, 0.957, 0.884, and 0.828, respectively (Fig. 4A). The model performed well in all seven testing sets, indicating that the “glmBoost+GBM” algorithm has a low risk of overfitting. Additionally, ROC values were plotted to individually diagnose NASH using the eight model genes in eight sets (Fig. 4B-I). Compared to FeRS, these eight model genes did not perform well in each cohort studied. Therefore, the FeRS is more suitable for diagnosing NASH in different cohorts. The model was compared with previously published NASH-related biomarkers. In total, 43 NASH diagnostic genes were identified in nine previous studies [24,25,26,27,28,29,30,31,32]. Notably, FeRS also exhibited good predictive performance (Fig. 5A-H). The combination of gene expression level differences across the eight cohorts showed that GRIA3, FADS2, FABP4, and PRKAA2 were upregulated in NASH, whereas NR4A1, IL6, ZFP36, and DHODH exhibited the opposite pattern (Fig. 6A-H).

Fig. 4
figure 4

A ferroptosis-related signature (FeRS) of nonalcoholic steatohepatitis (NASH) was developed. A The ROC values for diagnosing NASH using the 113 algorithms on the eight sets are shown in a heatmap. B-I Based on the eight model genes and FeRS, ROC values have been calculated for eight sets of NASH diagnoses

Full size image
Fig. 5
figure 5

AUC values are shown in bar graphs. A-H The ROC values for diagnosing NASH using FeRS and 51 genes across 8 cohorts were calculated and compared

Full size image
Fig. 6
figure 6

Gene expression histogram differences. A-H The 8 model genes were assessed for differences in expression levels. * P < 0.05, ** P < 0.01, *** P < 0.001

Full size image

Notably, model genes were closely associated with specific biological processes. Upregulation of FABP4 was implicated in “adherens junction” and “sodium reabsorption” (Fig. 7A). When PRKAA2 was upregulated, it was significantly enriched in “glycerolipid metabolism” and “steroid biosynthesis.” In addition, it is involved in the metabolism of various amino acids, including arginine, proline, histidine, and tyrosine (Fig. 7B). The upregulation of FADS2 and GRIA3 primarily involves the “TGF-β signaling pathway” associated with the regulation of inflammation (Fig. 7C, D). Furthermore, the upregulation of GRIA3 was significantly enriched in another inflammation-regulating pathway, the “JAK-STAT signaling pathway.” ZFP36 was significantly associated with both inflammation-regulating pathways (Fig. 7E). NR4A1 significantly enriched the pathways associated with “glycosphingolipid biosynthesis,” “circadian rhythms,” and “type 2 diabetes” (Fig. 7F). The downregulation of IL6 primarily enriched the pathways associated with “glycosphingolipid biosynthesis” (Fig. 7G). DHODH downregulation did not significantly enrich any biological pathways (Fig. 7H).

Fig. 7
figure 7

Enrichment pathway analysis. A-H The eight model genes were analyzed using GSVA

Full size image

Enrichment patterns in NASH

A better understanding of the pathogenesis of NASH was obtained by comparing the enriched biological pathways. “Cysteine and methionine metabolism,” “circadian rhythm,” “JAK-STAT signaling pathway,” “phenylalanine metabolism,” and “adipocytokine signaling pathway” exhibited significant upregulation in the NASH group (Fig. 8A).

Fig. 8
figure 8

Enrichment patterns of NASH. A GSVA enrichment analysis. B, C Patterns of immune landscape in patients with NASH. * P < 0.05, ** P < 0.01, *** P < 0.001

Full size image

This study further revealed whether the immune signatures and immune cells affect NASH (Fig. 8B and C). Notably, “type I IFN response,” “inflammation-promoting,” “cytolytic activity,” “T cells gamma delta,” “macrophages M0,” “macrophages M1,” “dendritic cells resting”, and “mast cells resting” were upregulated in NASH, while “CCR,” “B cells naive,” “plasma cells,” “dendritic cells activated,” and “neutrophils” exhibited the opposite trend.

Enrichment patterns for 8 model genes

An in-depth exploration of the biological significance of eight model genes was conducted. Twenty genes were closely associated with these eight genes (Fig. 9A). These 20 genes were significantly enriched in “PID AP1 pathway,” “PID P53 downstream pathway,” and “regulation of lipid metabolic processes” (Fig. 9B), as well as in the “MAPK family signaling cascades” associated with inflammation. Subsequently, their interactions were analyzed further in this study (Fig. 9C). ZFP36 ranked as the most important in FeRS, contrasting with FADS2 expression, yet showing synergy with IL6 expression (Fig. 9D). Notably, the downregulation of ZFP36 and IL6 suppressed the immune infiltration status of NASH, whereas the opposite was true for GRIA3 and FADS2 (Fig. 9E).

Fig. 9
figure 9

Model gene enrichment patterns. A PPI network map. B Biological pathway enrichment map. C Network map of gene correlation. D Friendship analysis. E Correlation network map. * P < 0.05, ** P < 0.01, *** P < 0.001

Full size image

Single cell analysis

A significant enrichment of these eight model genes was observed in the different single-cell datasets. The GSE159977(1) cohort was annotated using “SingleR” into 4 immune cell types (Fig. 10A). ZFP36 was significantly enriched in all four immune cell types, whereas NR4A1 was enriched only in NK cells, monocytes, and B cells (Fig. 10B and C). The remaining six genes were not significantly enriched. The GSE159977(2) cohort was similarly annotated using “SingleR” (Fig. 10D). By contrast, ZFP36 was significantly enriched in all four of the immune cell types, whereas NR4A1 was enriched only in monocytes (Fig. 10E and F). This suggests that ZFP36 was highly expressed in multiple immune cell types, whereas NR4A1 was primarily upregulated in monocytes. To validate this, we used the mouse single-cell data GSE129516, which was annotated using “SingleR” into 12 cell types (Fig. 10G). ZFP36 was significantly upregulated in B cells, dendritic cells, NK cells, and macrophages (Fig. 10H and I). There was an obvious enrichment of NR4A1 in NK cells and macrophages, whereas FABP4 was markedly enriched in NK cells, dendritic cells, and macrophages.

Fig. 10
figure 10

Single-cell annotations. a Four immune cells were annotated in GSE159977(1). b-c Enrichment landscapes of the model genes in immune cells. d Four immune cells were annotated in GSE159977(2).e-f Enrichment landscape of the model genes in immune cells. g Twelve immune cells were annotated in GSE129516. h-i Enrichment landscape of model genes in 12 types of cells

Full size image

Identification of translation regulators

Forty-four TFs and 23 miRNAs were identified (Fig. 11A-B), which likely influenced the development of NASH by regulating the translation of the model genes. In addition, 60 drugs targeting the five model genes were identified (Fig. 11C). This indicates the possibility of curing NASH.

Fig. 11
figure 11

Identification of TFs, miRNAs, and drugs. a Forty-four TFs regulate the translation process of the model genes. b Twenty-three miRNAs regulate the translation process of the model genes. c Predicted 60 therapeutic drugs for NASH

Full size image

Patterns of gene expression

The genes were validated at the mRNA level using clinical samples. qPCR revealed the upregulation of GRIA3 and PRKAA2 and the downregulation of DHODH in patients with NASH (Fig. 12a–c).

Fig. 12
figure 12

a-c GRIA3, DHODH, and PRKAA2 were differentially expressed. *** P < 0.001

Full size image

Discussion

To date, NASH has remained a challenge that has needed to be addressed [33]. Determining the drivers of inflammation is a major problem [34]. Because the liver is a major site for iron ion metabolism, ferroptosis is thought to be involved in the development of NASH [35]. Therefore, studies on ferroptosis may contribute to the management of NASH [36].

The 16 FRGs identified in this study are widely implicated in various metabolic pathways, such as “metabolism of alpha-linolenic acid,” “fat cell differentiation,” “regulation of fat cell differentiation,” and “circadian rhythm genes.” Notably, these genes have been linked to the causes of NASH, such as “fatty liver disease,” “dyslipidemias,” “insulin resistance,” “gestational diabetes,” and “unipolar depression” [37,38,39,40,41]. Sixteen FRGs were processed further. The “glmBoost+GBM” combination algorithm was optimal, which developed a FeRS with 8 genes. This predictive model diagnosed NASH with an ROC value higher than 0.80 in eight datasets. FeRS diagnosed NASH well in all eight datasets compared to other diagnostic markers for NASH, demonstrating the strong generalizability of the model.

A deeper examination of these eight model genes has revealed that ZFP36 ranked first in importance in patients with NASH and was significantly downregulated (P < 0.001). ZFP36 is implicated in the “JAK-STAT signaling pathway” [42] and “TGF-β signaling pathway,” which are closely associated with inflammation regulation [43]. ZFP36 is also associated with immune infiltration in NASH and its expression levels affect most immune signatures and cells. In particular, ZFP36 is positively associated with regulatory T cells and has anti-inflammatory effects. This implies that low ZFP36 expression in patients with NASH may promote the progression of inflammation by suppressing immune infiltration.

TFs can influence gene expression by regulating post-transcriptional translational processes [44]. Five ZFP36 TFs were identified in this study. This discovery facilitates further understanding of the specific mechanism of ZFP36 effects in NASH. Another microRNAs (miRNAs) that post-transcriptionally regulates gene expression was also analyzed [45]. It also inhibits translation and promotes RNA degradation [46]. Eight miRNAs bound to ZFP36 were identified. They may have essential functions in the promotion of NASH progression by ZFP36, affecting both metabolic and inflammatory pathways [47,48,49].

Additionally, the highly expressed genes GRIA3, FABP4, FADS2, and PRKAA2 have been implicated in lipid metabolism and liver fibrosis. They not only inhibit the anti-inflammatory effects of regulatory T cells but also promote the pro-inflammatory effects of M1 macrophages [50].In contrast, low expression of NR4A1, IL6, and DHODH involves key pathways such as “glycosphingolipid biosynthesis,” “circadian rhythms” and “type 2 diabetes,” activating pro-inflammatory M1 macrophages.

Study strengths and limitations

The advantage of this study is that compared to liver puncture biopsy, the FeRS not only can be applied conveniently and quickly on a large scale, but also avoids the great risks and pains that invasive operations bring to patients. However, the limitation is that NASH is a heterogeneous disease and the manifestations can vary according to individual differences. The FeRS may not be able to fully capture all potential variants, resulting in limited accuracy for different subtypes of NASH.

Conclusions

In this study, an FeRS was constructed and eight FRGs were identified. This model provides a noninvasive diagnostic method for NASH. Utilizing this noninvasive model helps to avoid complications associated with liver biopsy. Among the eight genes, ZFP36 ranked first. The downregulation of ZFP36 in NASH is not only involved in the “JAK-STAT pathway” and “TGF-βpathway” associated with inflammation regulation but also inhibits immune infiltration and promotes inflammation progression. The identification of ZFP36 has positive implications for the treatment of NASH, and its activation may inhibit NASH progression.

Availability of data and materials

Raw data were sourced from the GEO (https://www.ncbi.nlm.nih.gov/geo/) and FerrDb V2 (http://www.zhounan.org/ferrdb/current/) databases.

References

  1. Byrne CD, Targher G. NAFLD: a multisystem disease. J Hepatol. 2015;62:S47–64.

    Article  PubMed  Google Scholar 

  2. Zhou J, Zhou F, Wang W, Zhang XJ, Ji YX, Zhang P, et al. Epidemiological features of NAFLD from 1999 to 2018 in China. Hepatol. 2020;71:1851–64.

    Article  Google Scholar 

  3. Friedman SL, Neuschwander-Tetri BA, Rinella M, Sanyal AJ. Mechanisms of NAFLD development and therapeutic strategies. Nat Med. 2018;24:908–22.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  4. Bedossa P. Pathology of non-alcoholic fatty liver disease. Liver Int. 2017;37(Suppl 1):85–9.

    Article  PubMed  Google Scholar 

  5. Powell EE, Wong VW, Rinella M. Non-alcoholic fatty liver disease. Lancet. 2021;397:2212–24.

    Article  CAS  PubMed  Google Scholar 

  6. Loomba R, Friedman SL, Shulman GI. Mechanisms and disease consequences of nonalcoholic fatty liver disease. Cell. 2021;184:2537–64.

    Article  CAS  PubMed  Google Scholar 

  7. Brunt EM, Wong VW, Nobili V, Day CP, Sookoian S, Maher JJ, et al. Nonalcoholic fatty liver disease. Nat Rev Dis Primers. 2015;1:15080.

    Article  PubMed  Google Scholar 

  8. Wang XJ, Malhi H. Nonalcoholic fatty liver disease. Ann Intern Med. 2018;169:65–80.

    Google Scholar 

  9. Jiang X, Stockwell BR, Conrad M. Ferroptosis: mechanisms, biology and role in disease. Nat Rev Mol Cell Biol. 2021;22:266–82.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Chen J, Li X, Ge C, Min J, Wang F. The multifaceted role of ferroptosis in liver disease. Cell Death Differ. 2022;29:467–80.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Greener JG, Kandathil SM, Moffat L, Jones DT. A guide to machine learning for biologists. Nat Rev Mol Cell Biol. 2022;23:40–55.

    Article  CAS  PubMed  Google Scholar 

  12. Deo RC. Machine learning in medicine. Circulation. 2015;132:1920–30.

    Article  PubMed  PubMed Central  Google Scholar 

  13. Rauschert S, Raubenheimer K, Melton PE, Huang RC. Machine learning and clinical epigenetics: a review of challenges for diagnosis and classification. Clin Epigenetics. 2020;12:51.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Zhou N, Yuan X, Du Q, Zhang Z, Shi X, Bao J, et al. FerrDb V2: update of the manually curated database of ferroptosis regulators and ferroptosis-disease associations. Nucleic Acids Res. 2023;51:D571–d582.

    Article  CAS  PubMed  Google Scholar 

  15. Zhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10:1523.

    Article  PubMed  PubMed Central  Google Scholar 

  16. Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinform. 2013;14:7.

    Article  Google Scholar 

  17. Franz M, Rodriguez H, Lopes C, Zuberi K, Montojo J, Bader GD, et al. GeneMANIA update 2018. Nucleic Acids Res. 2018;46:W60–w64.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Duan Y, Ni S, Zhao K, Qian J, Hu X. Immune cell infiltration and the genes associated with ligamentum flavum hypertrophy: identification and validation. Front Cell Dev Biol. 2022;10:914781.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Xia J, Gill EE, Hancock RE. NetworkAnalyst for statistical, visual and network-based meta-analysis of gene expression data. Nat Protoc. 2015;10:823–44.

    Article  CAS  PubMed  Google Scholar 

  20. Castro-Mondragon JA, Riudavets-Puig R, Rauluseviciute I, Lemma RB, Turchi L, Blanc-Mathieu R, et al. JASPAR 2022: the 9th release of the open-access database of transcription factor binding profiles. Nucleic Acids Res. 2022;50:D165–d173.

    Article  CAS  PubMed  Google Scholar 

  21. Tastsoglou S, Skoufos G, Miliotis M, Karagkouni D, Koutsoukos I, Karavangeli A, et al. DIANA-miRPath v4.0: expanding target-based miRNA functional analysis in cell-type and tissue contexts. Nucleic Acids Res. 2023;51:W154–w159.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Wagner AH, Coffman AC, Ainscough BJ, Spies NC, Skidmore ZL, Campbell KM, et al. DGIdb 2.0: mining clinically relevant drug-gene interactions. Nucleic Acids Res. 2016;44:D1036–44.

    Article  CAS  PubMed  Google Scholar 

  23. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13:2498–504.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Baciu C, Pasini E, Angeli M, Schwenger K, Afrin J, Humar A, et al. Systematic integrative analysis of gene expression identifies HNF4A as the central gene in pathogenesis of non-alcoholic steatohepatitis. PLoS One. 2017;12:e0189223.

    Article  PubMed  PubMed Central  Google Scholar 

  25. He L, Wang J, Tao B, Zhu R, Li C, Ning B. Identification of ferroptosis-related genes in the progress of NASH. Front Endocrinol (Lausanne). 2023;14:1184280.

    Article  PubMed  PubMed Central  Google Scholar 

  26. Jiang ZY, Zhou Y, Zhou L, Li SW, Wang BM. Identification of key genes and immune infiltrate in nonalcoholic steatohepatitis: a Bioinformatic analysis. Biomed Res Int. 2021;2021:7561645.

    Article  PubMed  PubMed Central  Google Scholar 

  27. Liu MJ, Jin H, Chen YB, Yu JJ, Guo ZY, He SQ, et al. Screening of non-alcoholic steatohepatitis (NASH)-related datasets and identification of NASH-related genes. Int J Clin Exp Pathol. 2021;14:567–81.

    CAS  PubMed  PubMed Central  Google Scholar 

  28. Ma M, Xie W, Li X. Identification of autophagy-related genes in the progression from non-alcoholic fatty liver to non-alcoholic steatohepatitis. Int J Gen Med. 2021;14:3163–76.

    Article  PubMed  PubMed Central  Google Scholar 

  29. Meng Q, Li X, Xiong X. Identification of hub genes associated with non-alcoholic steatohepatitis using integrated bioinformatics analysis. Front Genet. 2022;13:872518.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Wang W, Liu X, Wei P, Ye F, Chen Y, Shi L, et al. SPP1 and CXCL9 promote non-alcoholic steatohepatitis progression based on bioinformatics analysis and experimental studies. Front Med (Lausanne). 2022;9:862278.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Yang S, Jia L, Xiang J, Yang G, Qiu S, Kang L, et al. KLF10 promotes nonalcoholic steatohepatitis progression through transcriptional activation of zDHHC7. EMBO Rep. 2022;23:e54229.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. Zhang JJ, Shen Y, Chen XY, Jiang ML, Yuan FH, Xie SL, et al. Integrative network-based analysis on multiple gene expression omnibus datasets identifies novel immune molecular markers implicated in non-alcoholic steatohepatitis. Front Endocrinol (Lausanne). 2023;14:1115890.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Lee KC, Wu PS, Lin HC. Pathogenesis and treatment of non-alcoholic steatohepatitis and its fibrosis. Clin Mol Hepatol. 2023;29:77–98.

    Article  PubMed  Google Scholar 

  34. Schuster S, Cabrera D, Arrese M, Feldstein AE. Triggering and resolution of inflammation in NASH. Nat Rev Gastroenterol Hepatol. 2018;15:349–64.

    Article  CAS  PubMed  Google Scholar 

  35. Wu J, Wang Y, Jiang R, Xue R, Yin X, Wu M, et al. Ferroptosis in liver disease: new insights into disease mechanisms. Cell Death Discov. 2021;7:276.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Tsurusaki S, Tsuchiya Y, Koumura T, Nakasone M, Sakamoto T, Matsuoka M, et al. Hepatic ferroptosis plays an important role as the trigger for initiating inflammation in nonalcoholic steatohepatitis. Cell Death Dis. 2019;10:449.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Polyzos SA, Kountouras J, Mantzoros CS. Obesity and nonalcoholic fatty liver disease: from pathophysiology to therapeutics. Metabolism. 2019;92:82–97.

    Article  CAS  PubMed  Google Scholar 

  38. Neuschwander-Tetri BA. Non-alcoholic fatty liver disease. BMC Med. 2017;15:45.

    Article  PubMed  PubMed Central  Google Scholar 

  39. Watt MJ, Miotto PM, De Nardo W, Montgomery MK. The liver as an endocrine organ-linking NAFLD and insulin resistance. Endocr Rev. 2019;40:1367–93.

    Article  PubMed  Google Scholar 

  40. Wu P, Wang Y, Ye Y, Yang X, Huang Y, Ye Y, et al. Liver biomarkers, lipid metabolites, and risk of gestational diabetes mellitus in a prospective study among Chinese pregnant women. BMC Med. 2023;21:150.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Chen D, Zhang Y, Huang T, Jia J. Depression and risk of gastrointestinal disorders: a comprehensive two-sample Mendelian randomization study of European ancestry. Psychol Med. 2023;53:7309–21.

    Article  PubMed  Google Scholar 

  42. Xu D, Yin C, Wang S, Xiao Y. JAK-STAT in lipid metabolism of adipocytes. Jakstat. 2013;2:e27203.

    PubMed  PubMed Central  Google Scholar 

  43. Fu X, Qie J, Fu Q, Chen J, Jin Y, Ding Z. miR-20a-5p/TGFBR2 Axis affects pro-inflammatory macrophages and aggravates liver fibrosis. Front Oncol. 2020;10:107.

    Article  PubMed  PubMed Central  Google Scholar 

  44. Spitz F, Furlong EE. Transcription factors: from enhancer binding to developmental control. Nat Rev Genet. 2012;13:613–26.

    Article  CAS  PubMed  Google Scholar 

  45. Saliminejad K, Khorram Khorshid HR, Soleymani Fard S, Ghaffari SH. An overview of microRNAs: biology, functions, therapeutics, and analysis methods. J Cell Physiol. 2019;234:5451–65.

    Article  CAS  PubMed  Google Scholar 

  46. Fabian MR, Sonenberg N, Filipowicz W. Regulation of mRNA translation and stability by microRNAs. Annu Rev Biochem. 2010;79:351–79.

    Article  CAS  PubMed  Google Scholar 

  47. Tan Y, Ge G, Pan T, Wen D, Gan J. A pilot study of serum microRNAs panel as potential biomarkers for diagnosis of nonalcoholic fatty liver disease. PLoS One. 2014;9:e105192.

    Article  PubMed  PubMed Central  Google Scholar 

  48. Wang TZ, Lin DD, Jin BX, Sun XY, Li N. Plasma microRNA: a novel non-invasive biomarker for HBV-associated liver fibrosis staging. Exp Ther Med. 2019;17:1919–29.

    CAS  PubMed  Google Scholar 

  49. Shi C, Huang F, Gu X, Zhang M, Wen J, Wang X, et al. Adipogenic miRNA and meta-signature miRNAs involved in human adipocyte differentiation and obesity. Oncotarget. 2016;7:40830–45.

    Article  PubMed  PubMed Central  Google Scholar 

  50. Tan RZ, Li JC, Zhu BW, Huang XR, Wang HL, Jia J, et al. Neuropeptide Y protects kidney from acute kidney injury by inactivating M1 macrophages via the Y1R-NF-κB-Mincle-dependent mechanism. Int J Biol Sci. 2023;19:521–36.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references

Funding

Anhui Provincial Department of Education (Project Number: 2023AH053301).

Author information

Author notes

  1. Longfei Dai, Wenkang Yuan and Renao Jiang contributed equally to this work.

Authors and Affiliations

  1. Department of General Surgery, The First Affiliated Hospital of Anhui Medical University, 218 Jixi Road, Hefei, 230022, Anhui Province, China

    Longfei Dai, Wenkang Yuan, Renao Jiang, Zhicheng Zhan, Liangliang Zhang, Xinjian Xu, Yuyang Qian, Wenqi Yang & Zhen Zhang

Authors
  1. Longfei Dai
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Wenkang Yuan
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Renao Jiang
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Zhicheng Zhan
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Liangliang Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Xinjian Xu
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Yuyang Qian
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Wenqi Yang
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Zhen Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

NASH data was downloaded, processed, and analyzed by LFD, WKY, RAJ. Experiments were carried out by ZCZ, LLZ, TY, XX, YYQ. The paper was corrected by WQY and ZZ. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Zhen Zhang.

Ethics declarations

Ethics approval and consent to participate

Not only did all participants sign an informed consent form before participation, but the study was approved by the Ethics Committee of the First Affiliated Hospital of Anhui Medical University (approval number: 2023497).

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Dai, L., Yuan, W., Jiang, R. et al. Machine learning-based integration identifies the ferroptosis hub genes in nonalcoholic steatohepatitis. Lipids Health Dis 23, 23 (2024). https://doi.org/10.1186/s12944-023-01988-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12944-023-01988-9

Keywords

Lipids in Health and Disease

ISSN: 1476-511X