Jan 04, 2024
Identification of Cancer-specific Constituent Elements inside Super-enhancers (cSEAdb)
- Xiang Liu1,
- Nancy Gillis1,
- Chang Jiang1,
- Anthony McCofie1,
- Timothy I Shaw1,
- Aik-Choon Tan2,
- Bo Zhao3,
- Lixin Wan1,
- Derek R Duckett1,
- Mingxiang Teng1
- 1Moffitt Cancer Center;
- 2Huntsman Cancer Institute, The University of Utah;
- 3Brigham and Women's Hospital and Harvard Medical School
Protocol Citation: Xiang Liu, Nancy Gillis, Chang Jiang, Anthony McCofie, Timothy I Shaw, Aik-Choon Tan, Bo Zhao, Lixin Wan, Derek R Duckett, Mingxiang Teng 2024. Identification of Cancer-specific Constituent Elements inside Super-enhancers (cSEAdb). protocols.io https://dx.doi.org/10.17504/protocols.io.kxygx38wzg8j/v1
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: January 03, 2024
Last Modified: January 04, 2024
Protocol Integer ID: 92932
Funders Acknowledgement:
NIH
Grant ID: R03DE030580
NIH
Grant ID: R01CA262530
NIH
Grant ID: R01CA255398
NIH
Grant ID: R01AI123420
Moffitt Cancer Center
Grant ID: P30CA076292
Disclaimer
DISCLAIMER – FOR INFORMATIONAL PURPOSES ONLY; USE AT YOUR OWN RISK
The protocol content here is for informational purposes only and does not constitute legal, medical, clinical, or safety advice, or otherwise; content added to protocols.io is not peer reviewed and may not have undergone a formal approval of any kind. Information presented in this protocol should not substitute for independent professional judgment, advice, diagnosis, or treatment. Any action you take or refrain from taking using or relying upon the information presented here is strictly at your own risk. You agree that neither the Company nor any of the authors, contributors, administrators, or anyone else associated with protocols.io, can be held responsible for your use of the information contained in or linked to this protocol or any of our Sites/Apps and Services.
Abstract
Super enhancers (SE) are large genomic elements composed of multiple constituent enhancers. As super enhancers are key regulators associated to cancer mechanisms, identifying cancer-specific super enhancer signatures improves our understanding of cancer-associated gene regulation. This protocol aims to provide a computational framework to identify cancer-specific super enhancer signatures at their constituent levels, using public H3L27Ac ChIP-seq data of the NCI-60 cancer cell panel. The protocol covers from data acquisition, pre-processing, statistical modeling and cancer-specific signature identification. It also provides links of scripts toward building an R data object for the storage, management and query of these signatures.
Enhancer and super-enhancer identification in each cancer cell line/sample
Enhancer and super-enhancer identification in each cancer cell line/sample
H3K27Ac ChIP-seq data acquisition
Download raw H3K27ac ChIP-seq data of NCI-60 human cancer cell lines from GEO
repositories with accession ID GSE143653. Fastq files of 60 cancer cell lines with two H3K27Ac replicates and one INPUT replicate should be downloaded from the SRA repository linked to this GEO accession.
CITATION
Enhancer and super-enhancer (SE) identification for each downloaded sample
Align raw ChIP-seq reads using Bowtie2
Align raw H3K27Ac ChIP-seq data of the NCI-60 cancer cell lines to human
genome hg38 using Bowtie2.
CITATION
Generate parallel bowtie2 alignment using the following script:
Use the command below to implement Bowtie2 parallel alignment for all ChIP-seq samples.
Command
GNU Parallel
Remove sequencing duplicate reads and sort the remaining reads
Use the following customized script to remove duplicated reads (minimizing potential PCR artifacts in the sequencing data) and sort sequencing reads (for downstream peak calling) in the aligned files.
Call peaks (enhancer candidates) of H3K27Ac data using MACS2
Call peaks based on the aligned BAM files using tool MACS2. The called peaks are treated as potential enhancer candidates for downstream analysis and for super enhancer identification.
CITATION
Generate parallel peak calling using the following script.
Use the command below to implement MACS2 parallel peak calling for all ChIP-seq samples.
Command
GNU Parallel
Convert file formats of MACS2 output for SE analysis
In order to call SEs on each ChIP-seq sample, some output files of MACS2 need to be re-formatted for fit SE calling software. Particularly, the ".narrowPeak" files from MACS2 have to be converted to ".gff" format using the following command.
Command
Convert .narrowPeak to .gff files for ROSE
SE identification using ROSE
For each cell line/sample, call super enhancers using the tool ROSE (https://hpc.nih.gov/apps/ROSE.html) based on files generated in the previous steps, including BAM files of H3K27Ac and INPUT, gff files and hg38 genome build.
CITATION
CITATION
Generate parallel super enhancer calling using the following script.
Use the command below to implement ROSE parallel peak calling for all ChIP-seq samples.
Command
GNU Parallel
Identification of cancer-specific super enhancer elements, documented in cSEAdb
Identification of cancer-specific super enhancer elements, documented in cSEAdb
Remove H3K27Ac ChIP-seq peaks overlapping with gene promotors and ChIP-seq blacklist regions
To avoid biases of promoter H3K27Ac signals, remove ChIP-seq peaks overlapping with gene promotors and blacklist regions using the following scripts. Promoter regions were defined as upstream 3000bp to downstream 1000bp surrounding genes' transcription start sites.
CITATION
Merge regions of different sample to generate a unique list of candidate enhancers and SEs
Merge enhancers and SEs across all cancer cell line samples using the scripts below. Basically, candidate enhancers and SEs from different samples might overlap with each other. The overlapped regions might indicate the same enhancer/SE regions across different cancer cells. We suggest only to merge consecutive regions with an overlap width at 25%.
CITATION
CITATION
Quantify ChIP-seq signals for each constituent enhancers (CE) inside all SEs
First, quantify ChIP-seq signals for all candidate enhancers using featureCount, including those inside SEs and normal enhancers. This will generate a signal matrix with rows as enhancers and columns as samples. Signals of the two replicates for each cancer cell line should be summed together. In other words, each cell line should be put as one column in the signal matrix. Second, signal matrix is normalized using the normalization method documented in DESeq2. Third, a subset of rows, representing only enhancers inside SE candidates are subtracted further for downstream cancer-specific analysis.
The customized code for the steps above is documented here.
CITATION
CITATION
LINK
https://doi.org/Imputation of Zero signals in the CE signal matrix
To ensure downstream mixture model fitting, zero signals in the signal matrix are imputed using the half of non-zero minimum signals in the signal matrix for each cell line . The following code can be used for the imputation. Here, se_ce_tmp is the signal matrix.
Command
Imputation of zero signals
Filter CEs with extreme low ChIP-seq coverage within their SEs
To ensure robust comparison of CEs across cancer cell lines, CEs with no more than 3% of coverage within their belonged SEs in any of the studied cell lines are excluded from downstream analysis. The filtering process is implemented with the following script. Basically, some rows (low signal values) of the signal matrix are removed.
Mixture model to identify active cell line from inactive cell line for each CE
Apply customized mixture model scrips (below) to model each CE. For each row of the signal matrix, the mixture model takes the 60 signals values from 60 cell lines and defines a probability for each value to be considered as active. In other words, the 60 instances of the same CE across 60 cell lines will be assigned as either active or inactive depending on whether the corresponding probability is larger or smaller than 0.5
Identify cancer-cell-specific CEs/SEs
Based on the frequency of being active across 60 cell lines for a given CE, cell-specific CEs are identified as less frequent but informative in encoding cancer cell identity. SEs holding these cell-specific CEs are concluded as cell-specific SEs.
In addition, the frequency of being inactive is also considered to identify cell-specific inactive CEs/SEs.
The below script is implemented to evaluate the frequency of active/inactive, the capability of encoding cell identity and the selection of cell-specific CEs and SEs.
Identify cancer-specific CEs/SEs
Cancer-specific CEs are identified based on the frequency of CEs being cell-specific for a given cancer type. In brief, if a cancer type has no less than 2 cell lines studied by the NCI-60 panel, a CE will be identified as cancer-specific only if it is cell-specific in at least 2 of the cell lines for this cancer type. For cancer types holding only on cell line, cell-specific CEs will be assigned as cancer-specific CEs.
SEs holding these cancer-specific CEs are concluded as cancer-specific SEs. In addition, cancer-specific inactive CEs/SEs are also identified.
The below script is implemented to identify cancer-specific SEs/CEs.
Create a data repository for cancer/cell-specific CEs/SEs - cSEAdb
To ease the use of the knowledge generated from the analysis, this protocol also provides scripts to generate R data object to store the cancer/cell-specific CEs and SEs.
The below script is implemented to generate such data object.
In addition, there is an R package (cSEAdb) available to explore, query and visualize the data object. For example, for a SE region, the cancer-specific CEs can be highlighted for a given cancer cell line or cancer type.
Citations
Step 2.1
Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2.
https://doi.org/10.1038/nmeth.1923Step 2.3
Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, Nusbaum C, Myers RM, Brown M, Li W, Liu XS. Model-based analysis of ChIP-Seq (MACS).
https://doi.org/10.1186/gb-2008-9-9-r137Step 2.5
Whyte WA, Orlando DA, Hnisz D, Abraham BJ, Lin CY, Kagey MH, Rahl PB, Lee TI, Young RA. Master transcription factors and mediator establish super-enhancers at key cell identity genes.
https://doi.org/10.1016/j.cell.2013.03.035Step 2.5
Lovén J, Hoke HA, Lin CY, Lau A, Orlando DA, Vakoc CR, Bradner JE, Lee TI, Young RA. Selective inhibition of tumor oncogenes by disruption of super-enhancers.
https://doi.org/10.1016/j.cell.2013.03.036Step 4
Sethi A, Gu M, Gumusgoz E, Chan L, Yan KK, Rozowsky J, Barozzi I, Afzal V, Akiyama JA, Plajzer-Frick I, Yan C, Novak CS, Kato M, Garvin TH, Pham Q, Harrington A, Mannion BJ, Lee EA, Fukuda-Yuzawa Y, Visel A, Dickel DE, Yip KY, Sutton R, Pennacchio LA, Gerstein M. Supervised enhancer prediction with epigenetic pattern recognition and targeted validation.
https://doi.org/10.1038/s41592-020-0907-8Step 4
Mantsoki A, Parussel K, Joshi A. Identification and Characterisation of Putative Enhancer Elements in Mouse Embryonic Stem Cells.
https://doi.org/10.1177/1177932220974623Step 5
Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features.
https://doi.org/10.1093/bioinformatics/btt656Step 5
Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2.
https://doi.org/