Analyze the data used:

The gene expression data (RNA-Seq, hg19) of HepG2 and hepatocyte cell lines were downloaded from ENCODE database.The ChIP-Seq data (hg19) for 11 histone modifications in HepG2 and hepatocyte were downloaded from ENCODE database. The RefSeq genes of human genome (hg19) were downloaded from the UCSC database.Clinical data and corresponding gene expression data of patients with liver hepatocellular carcinoma (LIHC) were downloaded from The Cancer Genome Atlas (TCGA) database.

The average histone modification signals of the different highly and lowly expressed genes

Tips:TH: the genes are highly expressed in tumor cell line but not in normal cell line; NH: the genes are highly expressed in normal cell line but not in tumor cell line; TL: the genes are lowly expressed in tumor cell line but not in normal cell line; NL: the genes are lowly expressed in normal cell line but not in tumor cell line;NH-TL: The genes are highly expressed in normal cell line and lowly expressed in tumor cell line;

Analysis result of figure above:

It was found that most of the histone modifications are stronger in highly expressed genes than that in lowly expressed genes by comparing the histone modification signals between TH (NH) and TL (NL) , suggesting that most of the histone modifications play an activating role in the two kinds of cell lines. However, H3K27me3 signals are stronger in lowly expressed genes than that in highly expressed genes for the two kinds of cell lines. These results indicate that the H3K27me3 may play an inhibiting role in the two kinds of cell lines. Moreover, we also observed that the signals of H3K27ac and H3K4me3 are stronger in tumor cell line than that in normal cell line. For the histone modifications of NH-TL, it can be seen that the H3K27me3 signals are stronger in tumor cell line than that in normal cell line. Furthermore, the H3K4me3 signals are stronger in tumor cell line.

The heat map of spearman’s correlations between histone modifications

Analysis result of figure above:

In highly expressed genes of tumor cell line, there are two histone modification clusters with strong positive correlation, including (H3K27me3, H3K9ac, H3K4me3, and H3K27ac) and (H4K20me1 and H3K36me3), there is just one histone modification cluster between H3K4me1 and (H3K27ac, H3K4me3, and H3K9ac) with strong negative correlation . For the lowly expressed genes of tumor cell line, there is only one histone modification cluster (H3K4me3, H3K4me2, H3K4me1, H2AFZ, and H3K9ac) that has strong positive correlation . Three histone modification clusters have strong positive correlation in highly expressed genes of normal cell line, including (H3K36me3, H3K79me2, and H4K20me1), (H2AFZ and H3K27ac) and (H3K4me2, H3K9ac and H3K4me3). In lowly expressed genes of normal cell line, there is just one histone modification cluster (H3K79me2, H3K36me3, and H3K9me3) that has strong positive correlation . However, the Spearman’s correlations between H3K4me2 and (H3K79me2, H3K36me3, and H3K9me3) have strong negative correlation.

Oncogenes and tumor suppressor genes known to be associated with hepatocellular carcinoma in the different highly and lowly expressed genes

Tips:DHLEG: different highly and lowly expressed genes. ONCO: oncogenes; TSG: tumor suppressor genes; TH: the genes are highly expressed in tumor cell line but not in normal cell line; TL: the genes are lowly expressed in tumor cell line but not in normal cell line; NH: the genes are highly expressed in normal cell line but not in tumor cell line; NL: the genes are lowly expressed in normal cell line but not in tumor cell line; NH-TL: the genes are highly expressed in normal cell line and lowly expressed in tumor cell line; NL-TH: the genes are lowly expressed in normal cell line and highly expressed in tumor cell line;

The histone modification signals in ONCO and TSG of the different highly and lowly expressed genes for two kinds of cell lines.

Tips:TH-ONCO: the highly expressed oncogenes in tumor cell line; NH-TSG: the highly expressed tumor suppressor genes in normal cell line;

Analysis result of figure above:

For tumor cell line, it shows that the average histone modification signals in highly expressed ONCO are generally stronger than that in lowly expressed TSG, especially for H3K4me3, H3K9ac, H3K27ac, and H3K4me2. Whereas for normal cell line, it shows that the average histone modification signals in highly expressed TSG are generally stronger than that in lowly expressed ONCO, except for H3K27me3.

The heat map of Spearman’s correlations between histone modification signals in the 80 bins of the promoter regions and the gene expression levels of TH-ONCO and NH-TSG.

Tips:Left:TH-ONCO;Right:NH-TSG;

Analysis result of figure above:

For the TH-ONCO in tumor cell line, there are the obvious negative correlations between gene expression levels and H3K27me3 in the regions (200, 250) bp and (850, 900) bp, while an obvious positive correlation between gene expression levels and H3K27ac in the region (1700, 1750) bp. For the NH-TSG in normal cell line, the heat map also shows that there are prominent correlations between gene expression levels and histone modifications, such as H3K9ac in the region (1500, 1550) bp, H3K4me3 in the regions (-1650, -1600) bp and (1800, 1850) bp. In addition, it has the obvious correlations between gene expression levels and H3K27ac in the regions (-1500, -1450), (850, 1000), (1050, 1100), (1150, 1250), (1300, 1450), and (1900, 1950) bp. The above results indicate that H3K4me3, H3K9ac, H3K27ac, and H3K27me3 are important histone modifications for gene expression.

Histone modification regions significantly correlated with gene expression levels

Key methylation sites and their genes in breast cancer

Analyze the data used:

The H3K4me3, H3K27me3, H3K79me3 and gene expression data in SRA format of CRC cell lines (DLD1: colorectal cancer cell line, SA: mesenchymal-like DLD1 derivative, METS: metastasis-derived DLD1 derivative) were downloaded from GEO. The reference number is GSE85688. It is worth noting that only three kinds of HMs data are available here. The RefSeq genes of the human genome (hg19) were downloaded from UCSC.

Partition of promoter regions:

Based on the RefSeq genes annotation file, the functional region of (-2000, 2000) bp flanking the TSS was divided into 40 bins of 100 bp in size for each gene.

Specification:

DLD1: colorectal cancer cell line; SA: mesenchymal-like DLD1 derivative; METS: metastasis-derived DLD1 derivative;EMT:DLD1 cells to SA cells;MET:SA cells to METS cells;

The Pearson correlation between the levels of H3K4me3, H3K27me3 and H3K79me3 (log2S ) and the level of gene expression ( log2E ) in DLD1, SA and METS, respectively

Tips:The r denotes the Pearson correlation coefficient between histone modification levels in promoters and gene expression levels;E represents the gene expression level;

Analysis result of figure above:

H3K4me3 and H3K79me3 were positively correlated with gene expression in DLD1, SA and METS cells, while H3K27me3 was negatively correlated with gene expression.

The top-ranked 20 important histone modification regions in the EU-MD genes and ED-MU genes respectively in colorectal cancer metastasis

Analyze the data used:

The genome-wide scale 11 HMs, 2 TFs, and polyAmRNAseq data in both human breast cancer cells (MCF-7) and human normal mammary epithelial cells (HMEC) were downloaded from the ENCODE.All data was annotated based on the Human reference genome hg38 (GRCh38). GRCh38 was downloaded from UCSC.

Division of flanking TSSs area:

Based on the RefSeq genes annotation file,the region of up- and down-stream 5 kb of the TSS for each RefSeq gene was divided into 100 bins (from 1-th to 100-th bins).

Specification:

HM: histone modifications; TF: transcription factor;

The signal distributions of H3K79me2, H3K27ac, and H3K4me1 in 100 bins flanking TSSs for MCF-7 and HMEC.

Tips:TSSs: transcription start sites;

Analysis result of figure above:

H3K79me2 levels in most bins of TSS downstream are increased for MCF-7.The levels of H3K27ac in MCF-7 are obviously increased in almost every bin flanking TSSs.H3K4me1 signal in each bin flanking TSSs is clearly enhanced for MCF-7.

The Spearman correlations between each HM/TF signals in each bin and the expression levels of the up-regulated genes and down-regulated genes in breast cancer cells (MCF-7).

Analysis result of figure above:

H3K27ac, H3K9ac, and H3K4me1/2/3 have stronger positive correlations with gene expression in MCF-7. However, the positive correlations between H3K79me2 and gene expression are obvious within TSS downstream . Moreover, around the 60-th bin, the correlations between H3K79me2 and gene expression are the strongest.

The signal distributions of 11 HMs and 2 TFs in MCF-7 and HMEC within the up- and downstream 5 kb of TSSs for the up-regulated genes.

Analysis result of figure above:

The change of H3K79me2 is especially obvious within TSS downstream.The signal changes of H3K27ac are obvious nearby TSSs (about the 40-th to 80-th bins).

The signal distributions of 11 HMs and 2 TFs in MCF-7 and HMEC within the up- and downstream 5 kb of TSSs for the down-regulated genes.

Analysis result of figure above:

The change of H3K79me2 is especially obvious within TSS downstream. Particularly, the difference of H3K79me2 levels between MCF-7 and HMEC around the 60-th bin is about 6 times. The signal changes of H3K27ac are obvious nearby TSSs (about the 40-th to 80-th bins), and H3K27ac level difference between the two types of cells is about 4.5 times. H3K4me1 changes evidently in almost 100 bins flanking TSSs. Moreover, the variations of three HMs are almost coincided with the changes of gene expression levels in MCF-7. The changes of H3K79me2, H3K27ac, and H3K4me1 may lead to the increase or decrease of gene expression in breast cancer.

The Spearman correlations between 11 HMs and 2 TFs in the up-regulated genes.

Analysis result of figure above:

In the up-regulated genes, H3K79me2 is positively correlated with H3K4me3. H3K4me3 is strong negatively correlated with H3K27me3.

The Spearman correlations between 11 HMs and 2 TFs in the down-regulated genes.

Analysis result of figure above:

For the down-regulated genes, there is a clear cluster of strong positively correlated HM pairs (H3K27ac, H3K9ac, and H3K4me2) ; H3K79me2 is strongly related to H3K4me2 and H3K27ac, respectively.

The top 10 significantly different up- and down-regulated genes in MCF-7

The top two important regions of each HM or TF, and HM or TF has been sorted in order of importance

The histone modification with the greatest importancein each bin

Analyze the data used:

The human genome location information (hg19) is downloaded from the UCSC database.The genome-wide profiles of 11 HMs and polyA plus RNA-seq data in GM12878 (B-lymphoblastoid cell, normal) and K562 (CML cell, cancer) are deposited in the ENCODE database.

Division of flanking TSSs area:

Based on the RefSeq genes annotation file,the region of up- and down-stream 5 kb of the TSS for each RefSeq gene was divided into 100 bins (from 1-th to 100-th bins).

Specification:

HM: histone modifications;

Statistical differences of histone modification signal levels between normal and chronic myelogenous leukemia cells across the 100 bins within the DNA regions flanking transcription start site (TSS).

Tips:Position 0 represents the TSS;

Analysis result of figure above:

Among the 11 HMs, except H2AFZ and H3K4me2, the rest of the HMs display remarkably dynamic changes in CML cells as compared with that in normal cells. The changes of H2AFZ signals mostly appear in the downstream regions of TSS, and H3K4me2 signal changes are significantly concentrated in the upstream regions of TSS.

The ratio of HM signals in K562 to that in GM12878 for down-differentially expressed genes

Analysis result of figure above:

For the down-DEGs, except H3K27me3, H3K4me3, and H4K20me1, the signals of other HMs reduce significantly in most bins within CML cells.

The ratio of HM signals in K562 to that in GM12878 for up-differentially expressed genes

Analysis result of figure above:

For the up-DEGs, except H3K27me3, the signals of other HMs are increased across all bins within CML cell lines.

Spearman correlation analysis of histone modification signal changes and gene expression changes for down-regulated differentially expressed genes

Analysis result of figure above:

The signals of H3K27me3 increase in all 100 bins within CML cells.The signals of H3K4me3 rise from the −15th to the 50th bins. The H4K20me1 signals are slightly enhanced from the −19th to the 50th bins.Cluster analysis exhibits that the impacts of HM signal changes on the expression level changes of downDEGs are divided into two categories. The first category includes H3K27me3, H3K9me3, and H4K20me1, while other HMs are classified into the second category . The increased signals of HMs in category 1 and the decreased signals of HMs within category 2 together lead to the down-regulation of gene expression.

Spearman correlation analysis of histone modification signal changes and gene expression changes for up-regulated differentially expressed genes

Analysis result of figure above:

The signals of H3K27me3 slightly decrease in all 100 bins within CML cells. Cluster analysis displays that the influences of HM signal changes on the expression changes of up-DEGs are also divided into two categories. The reduced signals of HMs (especially H3K27me3) in category 1 and the increased signals of HMs in category 2 together induce the up-regulation of gene expression levels, and those HMs in category 2 have stronger inducibility within the upstream regions of TSS.

The top 5 important regions of each HM, and HM has been sorted in order of importance

The rank for the significance of each HM across the 100 bins.

Analysis result of figure above:

It is noteworthy that H3K79me2 plays the most important role in almost all 100 bins, followed by H3K36me3. H3K27ac is relatively important for the regulation of gene expressions from the -20th to the 10th bins. H3K4me1 exerts its regulatory function from the 10th to the 50th bins. H3K4me3 has a crucial impact on gene expression changes in the -50th to -10th bins and the 20th–50th bins.

The most important histone modifications in each of the 100 bins.

Key gene in chronic myelogenous leukemia

Analyze the data used:

Human reference genome annotation data (GRCH38) are available in the UCSC Genome Browser.The genome-wide profiles of 11 HMs and polyA+ RNA-seq data for A549 cells (LUAD cells, tumor) and IMR90 cells (lung fibroblasts cells, normal) were indexed from the ENCODE database.Raw clinical profiles and RNA-seq data for patients with LUAD were retrieved from the TCGA dataset.

Division of flanking TSSs area:

Based on the RefSeq genes annotation file,the region of up- and down-stream 5 kb of the TSS for each RefSeq gene was divided into 100 bins (from 1-th to 100-th bins).

Specification:

HM: histone modifications;

Spearman correlation coefficients between expression changes of up-regulation differentially expressed genes and HM signal changes in each of the 100 bins

Analysis result of figure above:

For the up-regulation differentially expressed genes, the effects of HM signal changes on gene expression changes were stronger in the proximal regions of TSS than those in the distal regions.

Spearman correlation coefficients between expression changes of down-regulation differentially expressed genes and HM signal changes in each of the 100 bins

Analysis result of figure above:

For the down-regulation differentially expressed genes, HM signal changes in the distal regions of TSS contributed more to gene expression changes than those in the proximal regions.

The potential LUAD driver genes and the regions with clear H3K79me2 signal changes on these driver genes in lung adenocarcinoma

Analyze the data used:

The RNA-seq, DNA methylation, clinical and ATAC-seq data of human colon adenocarcinoma tissues were downloaded from The Cancer Genome Atlas database.The RefSeq genes annotation files of the Human reference genome hg38 (GRCh38) were got from UCSC.The interval of enhancer region was obtained from the FANTOM5 database. The annotation file of CGI region was downloaded from the UCSC database.

The distributions of average β values of each region in tumor and normal tissues.

Tips:HDGs-RC:The hypermethylated downregulated gene;HUGs-RC:The hypermethylated upregulated gene;

Analysis result of figure above:

The DNA methylation distributions in the two tissues were obviously different in HDGs-RC, especially the promoter regions (including TSS1500, TSS200, 5’UTR and 1stExon regions).Therefore, the hypermethylation in promoter regions is possible to lead to the downregulation of gene expression and inhibit gene expression. For the HUGs- RC, the hypermethylation of gene body and shore region are helpful to reinforce gene expression

Spearman's correlations between different DNA methylation modifications levels

Analysis result of figure above:

For down-regulated genes, there are four DNA methylation modification clusters that have strong positive correlation to inhibiting genes CD38, SEMA6D, IRF4 and IKZF1 expressions, respectively, (cg02183671, cg15994026, and cg26043257), (cg22389950, cg01053260, cg10864878, and cg17875483), (cg06392169, cg12741420, cg17228900, cg21277995, and cg06223767), and (cg16697214 and cg07589773) . For up-regulated genes, there is one DNA methylation modification cluster with strong positive correlation (cg04052466, cg26000619, cg26659805, and cg11325997), that promotes gene AMH expression .

Key methylation sites and their genes in colon adenocarcinoma

If you want to know more, please click here. More detailed information can be found in the "Detail" button on the search page.

Analyze the data used:

The DNA methylation data (Illumina Infinium Human Methylation 450K), gene expression data, and clinical data of human LIHC (liver hepatocellular carcinoma) tissue samples and adjacent normal tissue samples were downloaded from the TCGA database(hg38).The human reference genome annotation file (RefSeq, hg38) was downloaded from the UCSC database.The enhancer region location file of human was downloaded from the FANTOM5 database.The annotation file of the CGI region was downloaded from the UCSC database.

The 12 regions are further divided:

The promoter region was evenly divided into 30 bins, where each bin was 100 bp. The other 11 regions were normalized into 10 bins.

The density distribution of differentially methylated CpG sites in 12 different regions.

Tips:Hyper means hypermethylated;Hypo means hypomethylated;

Analysis result of figure above:

For the functional regions around the CGI, the distribution of differentially methylated CpG sites in the CGI region is significantly enriched, especially the edge regions on both sides of the CGI. In the genome regions (enhancer, promoter, 5'UTR, exon, intron, 3'UTR, and intergenic), the differentially methylated CpG sites are more abundant in the enhancer, promoter and exon regions, especially in the vicinity of the TSS (bin25). The distribution density of hypomethylated CpG sites is higher than that of hypermethylated CpG sites in the N_Shelf, N_Shore, S_Shore, and S_Shelf regions. In the edge regions on both sides of the CGI, the hypomethylated CpG sites are more abundant than the hypermethylated CpG sites. On the contrary, the distribution density of hypermethylated CpG sites is more than that of hypomethylated CpG sites in the interior of the CGI. The distribution density of hypomethylated CpG sites is greater than that of hypermethylated CpG sites in different regions of the genome.

The distribution of methylation levels for differentially methylated CpG sites in different regions of two tissues.

Tips:T means tumor tissues;N means normal tissues;

Key methylation sites and their genes in hepatocellular carcinoma

If you want to know more, please click here. More detailed information can be found in the "Detail" button on the search page.

Analyze the data used:

The gene expression data [fragments per kilobase of exon model per million mapped fragments (FPKM) and COUNTS], DNA methylation data (HM450K), and clinical data (hg38) of breast cancer and paracancerous tissues from the TCGA (The Cancer Genome Atlas) database. The human reference genome annotation file RefSeq gene (hg38) and the location file of CGI from UCSC. The position file of the enhancer from the FANTOM5 (Function Annotation of The Mammalian Genome) database.

The 12 regions are further divided:

The promoter region was evenly divided into 30 bins, where each bin was 100 bp. The other 11 regions were normalized into 10 bins.

The density distribution of DNA methylation probes in each functional region of the genome.

Tips:Hyper means hypermethylated;Hypo means hypomethylated;

Analysis result of figure above:

In both sides of CGI, intergenic regions, and intron regions, the degree of enrichment for hypomethylated probes is higher than that for hypermethylated probes. The hypermethylated probes are highly enriched in CGI and promoter regions. Then normalized the number of probes distributed in each region according to the length of each region. The enrichment degree of hypermethylated probes is higher than that of the hypomethylated probes in CGI, enhancer regions, and promoter regions. In other regions, the enrichment degree of hypomethylated probes is higher than that of hypermethylated probes. Hypermethylated probes are more significantly enriched in CGI and promoter regions after normalization by length. The hypermethylated probes in CGI and promoter regions are highly enriched both before and after normalization.Most of the hypermethylated probes in the promoter region are located on the CGI. The number of hypermethylated probes is almost the same as the number of hypomethylated probes in the promoter region without CGI. It can be speculated that the enrichment of hypermethylated probes in the promoter region is caused by the enrichment of CGI.

The enrichment number of DNA methylation probes in each region of up-regulated genes and down-regulated genes.

Analysis result of figure above:

The total number of ADMPs(abnormal DNA methylated probes) in up-regulated genes is less than that in down-regulated genes. The number of up-regulated genes is about 1.5 times that of down-regulated genes, which further shows that ADMPs like to be enriched in down-regulated genes. The number of probes enrichment in the 5′UTR and 3′UTR of the up-regulated genes is almost the same, and the same is true in the down-regulated genes. The number of hypermethylated probes in the promoter and exon regions of up-regulated genes is about twice the number of hypomethylated probes, and the same pattern is observed in the exons of down-regulated genes. However, the number of hypermethylated probes in the promoter region of down-regulated genes is 6.5 times that of hypomethylated probes.

Important genes that were highly correlated with clinical features

Key methylation sites and their genes in breast cancer

If you want to know more, please click here. More detailed information can be found in the "Detail" button on the search page.