Skip to main content
Download PDF

Volume 20 Supplement 2

Selected articles from the 17th Asia Pacific Bioinformatics Conference (APBC 2019): genomics

Genome-wide analysis of epigenetic dynamics across human developmental stages and tissues

BMC Genomics volume 20, Article number: 221 (2019) Cite this article

Abstract

Background

Epigenome is highly dynamic during the early stages of embryonic development. Epigenetic modifications provide the necessary regulation for lineage specification and enable the maintenance of cellular identity. Given the rapid accumulation of genome-wide epigenomic modification maps across cellular differentiation process, there is an urgent need to characterize epigenetic dynamics and reveal their impacts on differential gene regulation.

Methods

We proposed DiffEM, a computational method for differential analysis of epigenetic modifications and identified highly dynamic modification sites along cellular differentiation process. We applied this approach to investigating 6 epigenetic marks of 20 kinds of human early developmental stages and tissues, including hESCs, 4 hESC-derived lineages and 15 human primary tissues.

Results

We identified highly dynamic modification sites where different cell types exhibit distinctive modification patterns, and found that these highly dynamic sites enriched in the genes related to cellular development and differentiation. Further, to evaluate the effectiveness of our method, we correlated the dynamics scores of epigenetic modifications with the variance of gene expression, and compared the results of our method with those of the existing algorithms. The comparison results demonstrate the power of our method in evaluating the epigenetic dynamics and identifying highly dynamic regions along cell differentiation process.

Background

Lineage specification and maintenance of cellular identity are complex biological processes [1]. It is now widely accepted that cell phenotypes are significantly regulated by epigenetic states and that chromatin changes during differentiation contribute to the determination of cell fate [2]. Recent evidence further shows that coordinated epigenetic changes influence the maintenance of such cellular memory [3, 4]. DNA methylation and certain epigenetic modifications are essential for chromatin structures and gene expression in proper execution of developmental programs [5, 6]. Therefore, a fundamental question in the field is to exactly answer where and how the epigenetic changes regulate phenotypic changes.

To fully understand the dynamics and regulatory roles of epigenetic modifications, advanced sequencing technologies have generated genome-wide epigenetic maps of diverse developmental stages, lineages and tissues [7, 8]. In previous studies, researchers have differentiated human embryonic stem cells (hESCs) into mesendoderm, neural progenitor cells, trophoblast-like cells, and mesenchymal stem cells and systematically sequenced the transcriptome and epigenetic modifications of these lineages [9, 10]. The first three hESC derivatives reflects critical developmental linages in the embryo [11]. Mesenchymal stem cells have the ability of further multi-lineage differentiation to bone, cartilage, adipose, muscle, and connective tissues [12]. Mouse embryonic stem cells were also differentiated into a variety of precursor cell types [13]. The expanding body of epigenomic data permits researchers to study the dynamics of epigenetic marks. This is a key step to reveal regulatory roles of epigenetic modifications, and to understand how global features of epigenetic modifications impact cellular phenotypes across different developmental stages, lineages and tissues.

Most previous works focused on comparing the epigenetic modification profiles between two biological conditions, and further identifying regions that show differential patterns, such as ChIPDiff [14], diffReps [15], dPCA [16], HistoneHMM [17], csaw [18] and HMCan-diff [19]. While some other methods such as dMCA [20] and Yang’s method [21], were designed to detect cell-type-specific differential regions. Moreover, there are also some methods that were designed for identifying differential methylated region, such as QDMR [22] and MethylAction [23], whereas QDMR can also be applied to histone modification data analysis. Although several algorithms have been developed to analyze the epigenetic difference between two different conditions, little work devoting to differential analysis of epigenetic modifications among multiple cell types and across different developmental stages.

Here, we presented DiffEM, a computational method to quantify the dynamics of epigenetic marks and identified highly dynamic modification sites (HDMSs) across different human embryonic developmental stages. We applied this method to a public datasets with 6 intensely studied epigenetic marks of 20 different developmental stages and tissues. We identified HDMSs where different cell types exhibit distinctive epigenetic modification patterns, and found that these highly dynamic sites are enriched in genes related to cellular development and differentiation. We further correlated the dynamics scores of these epigenetic marks with those of gene expression levels. The results indicate that the changes of gene expression are closely related to the modification patterns of H3K4me1 and H3K27me3 in promoter regions during cell differentiation process. We compared DiffEM with the existing algorithms for identifying HDMSs. The comparison results show that DiffEM perform better in evaluating the epigenetic dynamics and identifying highly dynamic modification sites. This method is promising for broad applications in evaluating epigenetic dynamics in other complex biological processes.

Materials and methods

Datasets

To analyze the dynamic epigenetic changes during cellular differentiation and lineage specification, we obtained a large panel of epigenetic maps of human embryonic stem cells (hESCs) and the key derivatives, including trophoblast-like cells (TBL), mesendoderm (ME), neural progenitor cells (NPCs), and mesenchymal stem cells (MSCs). The iHMS [24] database has integrated massive genome-wide epigenetic modification maps and RNA expression data spanning different developmental stages and tissues. From iHMS, we downloaded 6 epigenetic modification maps (H3K4me1, H3K4me3, H3K9me3, H3K27ac, H3K27me3 and H3K36me3) of 20 different human developmental stages and tissues, including hESCs, the hESC-derived precursor cell types (TBL, ME, NPCs and MSCs), and 15 human primary tissues (adipose, adrenal gland, adult liver, aorta, esophagus, gastric, left ventricle, lung, ovary, pancreas, psoas muscle, right ventricle, right atrium, sigmoid colon, spleen, thymus, small intestine, breast, brain and bladder). Meanwhile, the RNA expression data and reference gene annotations were also downloaded from iHMS.

Overview of the DiffEM model

To characterize epigenetic dynamics across different development stages, we developed DiffEM, a new method to estimate the dynamics of epigenetic modifications based on hamming distance and identify highly dynamic modification sites. Unlike the previous work [20], we aimed to detect highly dynamic regions of epigenetic modification during cell differentiation process. To evaluate the dynamics across different differential stages and the primary tissues respectively, these 20 cell types were further categorized into three groups, hESC-derived precursor cell types, primary tissues and the whole group. We introduce the following steps to identify HDMSs, which are also shown in Fig. 1.

Fig. 1
figure 1

The flowchart of the DiffEM approach

Full size image

Data binarization. The raw ChIP-seq data were pre-processed in iHMS database [24]. The whole-genome was first segmented into 200bp bins. For each bin, neighboring read counts were summarized into an integer, indicating the extent of epigenetic modification in this region [25]. To reduce the effect of noise, we transformed these integers into binary values. First, we calculated the binarization threshold, by dividing the total read counts of all bins by the number of bins. If the read count of a bin is higher than the threshold, its binary value is set as 1, otherwise 0, After binarization, we noticed that some bins have no signals in all cell types, which may consist of sequences of low mappability. The consecutive regions with length more than 5 kb were removed from the genome. Finally, for the 6 investigated epigenetic marks, we obtained 6 binary matrices B K of size T (the number of cell types) by N (the number of 200 bp bins on the whole genome).

Calculation of the dynamics scores for each epigenetic mark. After data binarization, we calculated the dynamics scores for each epigenetic mark among multiple cell types. In particular, we used the hamming distance to measure the dynamics of each epigenetic modification. Here, we respectively calculated the dynamics scores of the 6 investigated epigenetic modifications in three cell type groups. As described above, given M cell types and N bins, we denoted b ktn as the binary profiles of epigenetic modification k for cell type t at position n. Then the difference between cell type t and others are calculated as:

$$ {Diff_{{ktn}}} ={\sum\nolimits}_{m=1,m\neq t}^{M} {hamming (b_{{ktn}}, b_{{kmn}})} $$
(1)

Further, the dynamics score of epigenetic modification k at position n was summed as:

$$ {DS_{{kn}}} ={\sum\nolimits}_{t=1}^{M} {Diff_{{ktn}}} $$
(2)

Identification of the highly dynamic modification sites. For each epigenetic mark, we have obtained the dynamics scores along the genome in each cell type group. The higher the dynamics score is, the greater the difference across these cell types exhibits. The sites with zero score were filtered first. Based on the calculated dynamics scores, we selected those bins whose dynamics scores are significantly higher than the genome background (p <0.05) and merged the adjacent bins into longer regions. These regions are referred to as highly dynamic modification sites (HDMSs).

Functional analysis of the highly dynamic modification sites

To investigate the potential functions of these identified HDMSs, we mapped them to RefSeq genes and some functional regions. According to their relative positions, we related the HDMSs to various genes when the centers of HDMSs are located in gene regions. The number of genes related to HDMSs was counted. Furthermore, we mapped the bins with the highest score to genomic features like promoter, coding region and exon. If a HDMS is not related to any gene, it is labeled as an intergenic sites. For further analysis of the functional relevance of HDMSs, we performed gene ontology (GO) enrichment analysis and pathway enrichment analysis for genes enriched with HDMSs via DAVID bioinformatics resources. The significant enrichment lists are obtained with p<0.05.

Comparisons among different epigenetic modifications

Epigenetic modifications play a critical role in cell differentiation process. Different epigenetic modifications may collaborate with each other to execute specific functions. We investigated the relations among different types of epigenetic marks. Based on the identified HDMSs of each epigenetic mark, the common HDMSs between different epigenetic modifications were obtained in the whole genome. Further, we estimated the correlations between the dynamics scores of these epigenetic modifications.

Correlation analysis between the dynamics of epigenetic mark and gene expression

First, we evaluated the dynamic scores of gene expression along the genome in each cell type group, which was calculated as the variance divided by the mean of gene expression. Then, we evaluated the correlation coefficients between the dynamic scores of epigenetic modifications and gene expression levels. For those identified HDMSs, a higher correlation coefficient indicates that gene expression is more easily regulated by the specific epigenetic modification.

Comparison among DiffEM, QDMR and IOD

As there exists no gold standard to benchmark highly dynamic modification sites, we adopted an indirect validation strategy. As previous studies [26], the validation was based on the correlations between the dynamics of epigenetic modifications and gene expression levels. To evaluate the performance in identifying HDMSs, we compared DiffEM with existing methods, QDMR and IOD. Unlike the methods restricted to the differential analysis between two cell types, the above three methods are capable of analyzing three or more cell types. QDMR was proposed for genome-wide differential analysis of epigenetic states based on Shannon entropy [22]. IOD was developed to detect differential regions across multiple cell types [27]. We first normalized the epigenetic data, and used QDMR and IOD to detect highly dynamic modification sites. These methods were compared by the correlations between the dynamics of epigenetic modifications and expression levels of the HDMSs.

Results

To investigate the dynamics of epigenetic modifications during cell differentiation process, we proposed a computational method, DiffEM, to quantify the dynamics score of various epigenetic marks and identify highly dynamic modification sites (HDMSs). We focused on human differentiation-related cell types, consisting of human embryonic stem cell, 4 hESC-derived precursor cell types, and 15 primary tissues. In each cell type, we collected 6 genome-wide epigenetic maps and gene expression datasets. DiffEM was applied to identify HDMSs along cell differentiation process. To evaluate the performance of our proposed method, in this section we analyzed the identified HDMSs to discover their potential biological roles during cell differentiation and development. Furthermore, we compared DiffEM with two previous methods, QDMR and IOD.

Genome-wide characterization of epigenetic modification dynamics

To better explore the dynamic epigenetic changes across different cell differentiation stages, these 20 cell types were further grouped into three groups, hESCs and hESC-derived precursor cell types, primary tissues and the whole group. For each group and each epigenetic modification mark, we quantified the dynamics score for each bin based on hamming distance, and then ranked these bins according to their dynamics scores. We selected those bins whose dynamics scores were significantly higher than the genome background (p <0.05).

After merging the neighboring bins, we obtained the HDMSs for each epigenetic modification in each group. For different epigenetic marks, we found that there exist big overlaps between the HDMSs of different epigenetic modifications. This is consistent with previous finding that the epigenetic modifications collaborated with each other to consummate certain regulatory function. As shown in Fig. 2, we respectively calculated the percentage of overlapping HDMSs among 6 epigenetic modifications in these three groups. On the whole, the overlapping sites make up 20%˜60% of total HDMSs in different groups. In the hESCs and hESC-derived precursor group, the HDMSs of different epigenetic marks overlap more than those of the other two groups. For example, the overlap rates of H3K4me1 with other five epigenetic marks range from 40% to 50% in hESC-derived group, while those overlap rates in the other two groups are not greater than 25%. Specifically, H3K4me3 is highly overlapped with H3K9me3 and H3K27ac. These observations demonstrate that epigenetic modifications collaborate closely to regulate the cell differentiation process [4].

Fig. 2
figure 2

The overlaps between the top HDMSs of different epigenetic marks. The value in row i column j represents the proportion of HDMSs of epigenetic modification i overlapped by those of epigenetic modification j. a The hESCs and hESC-derived precursor group. b The primary tissues group. c The whole group

Full size image

As distinct epigenetic modifications share HDMSs, we further investigated the correlation between the dynamics scores of different epigenetic marks. As shown in Fig. 3, the investigated epigenetic marks demonstrate varied correlation in the three comparison groups. In particular, the epigenetic marks show higher correlation in the hESCs and hESC-derived precursor group. This result indicates that the dynamics of epigenetic modifications are similar during the cell differentiation process, which is conformed to the results of previous overlaps analysis.

Fig. 3
figure 3

The heatmaps representing the correlations between each pair of epigenetic modifications. a The hESCs and hESC-derived precursor group. b The primary tissues group. c The whole group

Full size image

Highly dynamic modification sites are related to various genomic features

Further, we mapped the identified HDMSs to RefSeq genes and collected the genes enriched with HDMSs for each epigenetic mark. Here we explored how the dynamic epigenome participates in early embryonic developmental stages and focused on the hESCs and hESC-derived precursor group. To examine the potential functions of those genes, we performed systematic gene ontology enrichment analysis using DAVID tools (https://david.ncifcrf.gov/) and summarized the key biological processes and pathways for each epigenetic mark. Overall, for the aforementioned six epigenetic modification marks, we found that those HDMSs-enriched genes exhibit enrichment for cell differentiation and development functions (Table 1) (p value <0.05). For example, GO terms related to development such as ’nervous system development’ are enriched in HDMSs of H3K4me1, H3K4me3, H3K9me3,H3K27ac, H3K27me3, GO terms related to differentiation such as ’neuron differentiation’ and ’cerebellar granule cell differentiation’ are enriched in HDMSs of H3K4me1, H3K9me3, H3K27me3, H3K36me3. This is consistent with previous finding that regulatory elements essential for cellular identity are often epigenetically modified in parental cells [28, 29]. The results highlight the importance of stage-specific epigenetic modification patterns of transcription factors for defining the developmental potentials.

Table 1 Functional enrichment of genes on the whole genome of six histone modifications
Full size table

Also, we noticed that the biological processes of distinct epigenetic marks have overlappings. One possible interpretation for this observation could be that these epigenetic marks may have the same changing trend, collaborating with each other to finish the complex regulatory functions. Taken together, the above results of GO annotation demonstrated the power of our method in identifying the highly dynamic sites of these epigenetic modifications. And, the results strongly suggest that the HDMSs mark critical regulatory regions for cell differentiation and development process. Further characterization of epigenetic modification patterns and gene expression within HDMSs may provide important insights into the regulatory functions of the specific epigenetic patterns.

Highly dynamic modified sites neighboring genes reveal diverse transcriptional patterns

To analyze the regulatory roles of these dynamic epigenetic patterns, we further explored the epigenetic modification and gene expression patterns within HDMSs. We computed the correlation coefficients between the dynamics of epigenetic modifications and gene expression levels of the HDMSs-enriched genes. We mapped the HDMSs to Ref-Seq genes and obtained gene expression of the associated genes. As these 20 cell types were divided into three groups, the dynamics score of gene expression was assessed using the same method as epigenetic marks (see Methods). For those HDMSs located in promoters, and coding regions, the Pearson correlation coefficients were respectively computed.

As shown in Fig. 4, we noted that there is highly correlation between the dynamics of gene expression level and epigenetic modification in promoter regions. Relatively, the correlation in coding regions is lower. These results indicate that the variance of epigenetic modification patterns in promoter regions has a higher regulatory role than that in coding regions. The three different groups have a similar trend. In detail, the six epigenetic modification marks exhibit different regulatory effect. For the hESCs and hESC-derived precursor group, the dynamics of gene expression levels are highly regulated by the modification patterns of H3K4me1 and H3K27me3 in promoter regions. For the primary tissues, the correlations are much higher for H3K9me3 and H3k27ac.

Fig. 4
figure 4

The correlations between the dynamics of epigenetic modification and that of expression level in HDMSs. a The hESCs and hESC-derived precursor group. b The primary tissues group. c The whole group

Full size image

Comparison with QDMR and IOD in identifying HDMSs

Considering that our method was developed for the differential analysis for multiple cell types, we compared DiffEM with two similar previous methods QDMR and IOD [22, 27], which were also designed for multiple conditions. QDMR is based on Shannon entropy [22], and IOD is defined as the variance divided by the mean value [27]. The performance was measured by the correlation analysis between the epigenetic modification dynamics and gene expression difference.

Firstly, we respectively identified the highly dynamic modification sites using these three methods, and ranked the HDMSs according to the dynamics score. Similarly, we obtained the ranked highly dynamic expression sites. Then, we associated these HDMSs with the highly dynamic expression sites by bitwise matching. To evaluate the performance of these three methods, we define two metrics, MatchedNum and AveDS. MatchedNum is computed as the number of highly dynamic expression sites matching with the top ranked HDMSs, which is similar to recall. AveDS represents the average dynamics score of these matched highly dynamic expression sites. Here, for fair comparison among the three methods, we calculated the entropy as the average dynamics score as QMDR.

We compared the performance on the aforemetioned 6 epigenetic modifications, the results are shown in Fig. 5 and Additional file 1. Figure 5 shows the comparison results for the hESCs and hESC-derived precursor group. Figure S1, Figure S2 (see Additional file 1) showed the results of the other two groups. We first compared the matched numbers of all differential gene expression sites output by these methods. Our method could get a higher MatchedNum of highly dynamic expression sites than those of QDMR and IOD (Fig. 5a). However, this raises the question that to what extent these matched sites are dynamically expressed. As we noted that changes in epigenetic modifications could cause differential expression of related genes, we further compared the average dynamics of gene expression of these matched sites. Lower ave indicate better performance. As the results showed (Fig. 5b), our method has good performance in AveDS. These observations demonstrate that our method always achieves a balance between matched MatchedNum and AveDS, which means our approach could be applied to find meaningful HDMSs as many as possible. In addition, the overall analysis for MatchedNum and AveDS shows that IOD may be applicable to detecting the highest HDMSs, because of the commonly small Num but better AveDS of related differential gene expression sites. In summary, our method outperforms the two existing methods in identifying the HDMSs across different developmental stages and tissues in the whole genome.

Fig. 5
figure 5

Performance comparisons among our proposed method, IOD and QDMR for each epigenetic mark. The performance was evaluated by MatchedNum and AveDs. The higher the Num is, and the lower the Ave is, the better the performance in detecting HDMSs. (a) The MatchedNum, (b) The AveDss

Full size image

Discussion

In this paper, we proposed a new computational method, DiffEM, based on hamming distance to identify the highly dynamic modification sites that undergo chromatin changes during human cell differentiation process. Different from previous methods that mostly focused on differential analysis between two cell types, our method is designed for differential analysis of genome-wide epigenetic modification across multiple cell types. DiffEM can be broadly applied in a range of studies involving various epigenetic marks in different conditions. We applied this approach to investigating 6 epigenetic marks of 20 human cell types, including hESCs, 4 hESC-derived Lineages and 15 human primary tissues. We identified highly dynamic modification sites where different cell types exhibit distinctive epigenetic modification patterns, and found that these highly dynamic modification sites are enriched in the genes are related to cellular development and differentiation. The results also demonstrate the strong association among the dynamics of different epigenetic marks, consistent with previous finding that different epigenetic modifications collaborate with each other to consummate complex regulatory functions. Further, we evaluated the effectiveness of our method, by correlating the dynamics scores of epigenetic modification with the variance of gene expression. We compared DiffEM with two existing methods, QDMR and IOD. The comparison results indicate the power of our method in quantifying the epigenetic dynamics and identifying highly dynamic regions.

Abbreviations

DAVID:

The database for annotation, visualization and integration discovery

GO:

Gene ontology

HDMSs:

Highly dynamic modification sites

hESCs:

Human embryonic stem cells

iHMS:

A database integrating human histone modification data across developmental stages and tissues

ME:

Mesendoderm

MSCs:

Mesenchymal stem cells

NPCs:

Neural progenitor cells

TBL:

Trophoblast-like cells

References

  1. Atlasi Y, Stunnenberg HG. The interplay of epigenetic marks during stem cell differentiation and development. Nat Rev Genet. 2017; 18(6990):643–58.

    Article  CAS  Google Scholar 

  2. Kouzarides T. Chromatin modifications and their function. Cell. 2007; 128:693–705.

    Article  CAS  Google Scholar 

  3. Rivera CM, Ren B. Mapping human epigenomes. Cell. 2013; 155(1):39–55.

    Article  CAS  Google Scholar 

  4. Dixon JR, Jung I, Selvaraj S, Shen Y, Antosiewicz-Bourget JE, Lee AY, et al. Chromatin architecture reorganization during stem cell differentiation. Nature. 2015; 518(7539):331–6.

    Article  CAS  Google Scholar 

  5. Mikkelsen TS, Ku M, Jaffe DB, Issac B, Lieberman E, Giannoukos G, Alvarez P, Brockman W, Kim TK, Koche RP. Genome-wide maps of chromatin state in pluripotent and lineage-committed cells. Nature. 2007; 448(7153):553.

    Article  CAS  Google Scholar 

  6. Tsankov AM, et al. Transcription factor binding dynamics during human es cell differentiation. Nature. 2015; 518(7539):344–9.

    Article  CAS  Google Scholar 

  7. Ernst J, Kheradpour P, Mikkelsen TS, Shoresh N, Ward LD, Epstein CB, Zhang X, Wang L, Issner R, Coyne M, et al. Mapping and analysis of chromatin state dynamics in nine human cell types. Nature. 2011; 473(7345):43–9.

    Article  CAS  Google Scholar 

  8. Roadmap EC, Kundaje A, Meuleman W, Ernst J, et al. Integrative analysis of 111 reference human epigenomes. Nature. 2015; 518(7539):317–30.

    Article  Google Scholar 

  9. Gifford CA, Ziller MJ, Gu H, Trapnell C, Donaghey J, Tsankov A, Shalek AK, Kelley DR, Shishkin AA, Issner R, et al. Transcriptional and epigenetic dynamics during specification of human embryonic stem cells. Cell. 2013; 153(5):1149–63.

    Article  CAS  Google Scholar 

  10. Xie W, Schultz MD, Lister R, et al. Epigenomic analysis of multilineage differentiation of human embryonic stem cells. Cell. 2013; 153(5):1134–48.

    Article  CAS  Google Scholar 

  11. Thomson JA, Itskovitz-Eldor J, Shapiro SS, Waknitz MA, Swiergiel JJ, Marshall VS, Jones JM. Embryonic stem cell lines derived from human blastocysts. Science. 1998; 282(5391):1145.

    Article  CAS  Google Scholar 

  12. Vodyanik MA, Yu J, Zhang X, Tian S, Stewart R, Thomson JA, Slukvin II. A mesoderm-derived precursor for mesenchymal stem and endothelial cells. Cell Stem Cell. 2010; 7(6):718.

    Article  CAS  Google Scholar 

  13. Yu P, Xiao S, Xin X, Song C-X, Huang W, McDee D, Tanaka T, Wang T, He C, Zhong S. Spatiotemporal clustering of the epigenome reveals rules of dynamic gene regulation. Genome Res. 2013; 23(2):352–64.

    Article  CAS  Google Scholar 

  14. Xu H, Wei CL, Lin F, Sung WK. An hmm approach to genome-wide identification of differential histone modification sites from chip-seq data. Bioinformatics. 2008; 24(20):2344–9.

    Article  CAS  Google Scholar 

  15. Shen L, Shao NY, Liu X, Maze I, Feng J, Nestler EJ. diffreps: Detecting differential chromatin modification sites from chip-seq data with biological replicates. PloS ONE. 2013; 8(6):65598.

    Article  Google Scholar 

  16. Ji H, Li X, Wang QF, Ning Y. Differential principal component analysis of chip-seq. Proc Natl Acad Sci U S A. 2013; 110(17):6789.

    Article  CAS  Google Scholar 

  17. Heinig M, Colom-Tatch M, Taudt A, Rintisch C, Schafer S, Pravenec M, Hubner N, Vingron M, Johannes F. histonehmm: Differential analysis of histone modifications with broad genomic footprints. BMC Bioinformatics. 2015; 16(1):1–15.

    Article  Google Scholar 

  18. Lun ATL, Smyth GK. csaw: a bioconductor package for differential binding analysis of chip-seq data using sliding windows:. Nucleic Acids Res. 2016; 44(5):45.

    Article  Google Scholar 

  19. Ashoor H, Louis-Brennetot C, Janoueix-Lerosey I, Bajic VB, Boeva V. Hmcan-diff: a method to detect changes in histone modifications in cells with different genetic characteristics. Nucleic Acids Res. 2017; 45(8):58.

    Google Scholar 

  20. Chen C, Zhang S, Zhang XS. Discovery of cell-type specific regulatory elements in the human genome using differential chromatin modification analysis. Nucleic Acids Res. 2013; 41(20):9230.

    Article  CAS  Google Scholar 

  21. Yang X, Shao X, Gao L, Zhang S. Systematic dna methylation analysis of multiple cell lines reveals common and specific patterns within and across tissues of origin. Hum Mol Genet. 2015; 24(15):4374.

    Article  CAS  Google Scholar 

  22. Zhang Y, Liu H, Lv J, Xiao X, Zhu J, Liu X, Su J, Li X, Wu Q, Wang F. Qdmr: a quantitative method for identification of differentially methylated regions by entropy. Nucleic Acids Res. 2011; 39(9):58.

    Article  Google Scholar 

  23. Bhasin JM, Bo H, Ting AH. Methylaction: detecting differentially methylated regions that distinguish biological subtypes. Nucleic Acids Res. 2016; 44(1):106–16.

    Article  CAS  Google Scholar 

  24. Gan Y, Han T, Guan J, Zhou S. ihms: a database integrating human histone modification data across developmental stages and tissues. BMC Bioinformatics. 2017; 18(1):103.

    Article  Google Scholar 

  25. Gan Y, Tao H, Zou G, Yan C, Guan J. Dynamic epigenetic mode analysis using spatial temporal clustering. BMC bioinformatics. 2016; 17(17):537.

    Article  Google Scholar 

  26. Zhang Y, An L, Yue F, Hardison RC. Jointly characterizing epigenetic dynamics across multiple human cell types. Nucleic Acids Res. 2016; 44(14):6721–31.

    Article  CAS  Google Scholar 

  27. Pinello L, Xu J, Orkin SH, Yuan GC. Analysis of chromatin-state plasticity identifies cell-type-specific regulators of h3k27me3 patterns. Proc Natl Acad Sci U S A. 2014; 111(3):344.

    Article  Google Scholar 

  28. Wang H, Liu C, Deng L. Enhanced prediction of hot spots at protein-protein interfaces using extreme gradient boosting. Sci Rep. 2018; 8(1):14285.

    Article  Google Scholar 

  29. Pan Y, Liu D, Deng L. Accurate prediction of functional effects for variants by combining gradient tree boosting with optimal neighborhood properties. PloS ONE. 2017; 12(6):0179314.

    Article  Google Scholar 

Download references

Acknowledgements

Authors are grateful to NIH Roadmap Epigenome Project and iHMS website for providing the epigenomic data to carry out this work.

Funding

This work and the publication costs were supported in part by the Fundamental Research Funds for the Central Universities (2232016A3-05), the National Natural Science Foundation of China (61772128, 61772367), National Key Research and Development Program of China (2016YFC0901704) and Shanghai Natural Science Foundation (17ZR1400200,18ZR1414400).

Availability of data and material

Supplementary data are available at Github: https://github.com/xiazhang0513/DiffEM.

About this supplement

This article has been published as part of BMC Genomics Volume 20 Supplement 2, 2019: Selected articles from the 17th Asia Pacific Bioinformatics Conference (APBC 2019): genomics. The full contents of the supplement are available online at https://bmcgenomics.biomedcentral.com/articles/supplements/volume-20-supplement-2.

Author information

Author notes

  1. Xia Zhang and Yanglan Gan contribute equally to the paper.

Authors and Affiliations

  1. School of Computer Engineering and Science, Shanghai University, Shanghai, China

    Xia Zhang & Guobing Zou

  2. School of Computer Science and Technology, Donghua University, Shanghai, China

    Yanglan Gan

  3. Department of Computer Science and Technology,Tongji University, Shanghai, China

    Jihong Guan

  4. Shanghai Key Lab of Intelligent Information Processing, and School of Computer Science, Fudan University, Shanghai, China

    Shuigeng Zhou

Authors
  1. Xia Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Yanglan Gan
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Guobing Zou
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Jihong Guan
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Shuigeng Zhou
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

YLG and XZ are responsible for the main idea, as well as the completion of the manuscript and experiment. GBZ, SGZ and JHG have coordinated data preprocessing and supervised the effort. All authors have read and approved the final manuscript.

Corresponding author

Correspondence to Yanglan Gan.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Publisher’s Note

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

Additional file

Additional file 1

Figure S1. Performance comparisons for primary tissues among our method DiffEM, IOD and QDMR for each epigenetic mark. (A) The MatchedNum, (B) The AveDS. Figure S2. Performance comparisons for the whole group among our method DiffEM, IOD and QDMR for each epigenetic mark. (A) The MatchedNum, (B) The AveDS. (PDF 405 kb)

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. 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.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhang, X., Gan, Y., Zou, G. et al. Genome-wide analysis of epigenetic dynamics across human developmental stages and tissues. BMC Genomics 20 (Suppl 2), 221 (2019). https://doi.org/10.1186/s12864-019-5472-0

Download citation

  • Published:

  • DOI: https://doi.org/10.1186/s12864-019-5472-0

Keywords

BMC Genomics

ISSN: 1471-2164

Contact us