Updated: Nov-12-2024
Overview
The MotifPeeker package facilitates the comparison and validation of datasets from epigenomic profiling methods, using motif enrichment as the key benchmark. The package generates a comprehensive summary report with results from various downstream analyses by processing peak, alignment, and motif files. This allows for detailed statistical analysis of multiple epigenomic datasets without any coding, ensuring both accessibility and robustness.
Introduction
The rapidly advancing field of epigenomics has led to the development of various techniques for profiling protein interactions with DNA, enhancing our understanding of gene regulatory mechanisms and genetic factors behind complex diseases. However, the validation of these newer methods, such as CUT&RUN, CUT&TAG and TIP-Seq, remains a critical area that requires further exploration, especially given their potential to address the challenges of traditional ChIP-Seq.
Common epigenomic profiling techniques rely on target proteins, such as the transcriptional regulator CTCF, binding to their respective sites on the DNA to isolate the sequences for sequencing. These binding sites may contain specific sequences recognised by the transcription factors, called motifs. Unlike other comparison tools like ChIPseeker and EpiCompare, MotifPeeker checks for the presence of these motifs in the sequences enriched from epigenomic profiling methods as a novel strategy to benchmark them.
At the same time, general metrics like FRiP scores and peak width distributions are also reported to add more context to the comparisons. While the goal remains to benchmark different epigenomic datasets, MotifPeeker can also be used to compare the effects of various downstream processing, such as the thresholds for peak calling and the choice of the peak caller itself. The package can also help identify differences arising from different experimental conditions or protocol optimisations.
Data
MotifPeeker comes with a small subset of two epigenomic datasets targeting CTCF in HCT116 cells, generated using ChIP-Seq and TIP-Seq.
- ChIP-Seq alignment file (
CTCF_ChIP_alignment.bam) sourced from the ENCODE project (Accession: ENCFF091ODJ).
- TIP-Seq alignment file (
CTCF_TIP_alignment.bam) was manually processed using thenf-core/cutandrunpipeline. The raw read files were sourced from NIH Sequence Read Archives (ID: SRR16963166).
The alignment files were processed using the MACS3 peak
caller to produce their respective peak files with the
q-value parameter set to 0.01.
Two motif files for CTCF are also bundled with the package:
Please note that the peaks and alignments included are a very small subset (chr10:65,654,529-74,841,155) of the actual data. It only serves as an example to demonstrate the package and run tests to maintain the integrity of the package.
Installation
MotifPeeker uses memes
which relies on a local install of the MEME suite, which can be
installed as follows:
MEME_VERSION=5.5.5 # or the latest version
wget https://meme-suite.org/meme/meme-software/$MEME_VERSION/meme-$MEME_VERSION.tar.gz
tar zxf meme-$MEME_VERSION.tar.gz
cd meme-$MEME_VERSION
./configure --prefix=$HOME/meme --with-url=http://meme-suite.org/ \
--enable-build-libxml2 --enable-build-libxslt
make
make install
# Add to PATH
echo 'export PATH=$HOME/meme/bin:$HOME/meme/libexec/meme-$MEME_VERSION:$PATH' >> ~/.bashrc
echo 'export MEME_BIN=$HOME/meme/bin' >> ~/.bashrc
source ~/.bashrcNOTE: It is important that Perl dependencies
associated with MEME suite are also installed, particularly
XML::Parser, which can be installed using the following
command in the terminal:
For more information, refer to the Perl dependency section of the MEME suite.
Once the MEME suite and its associated Perl dependencies are
installed, install and load MotifPeeker:
Alternatively, you can use the Docker/Singularity container to run the package out-of-the-box.
Running MotifPeeker
In this example, we will compare the bundled ChIP-Seq dataset against the TIP-Seq dataset.
Prepare input data
Peak Files
MotifPeeker accepts lists of both GRanges
objects produced by read_peak_file(), or paths to the
MACS2/3 .narrowPeak files or SEACR
.bed files, or ENCODE file IDs to automatically download
the respective files.
## MACS2/3 peak files
peak_files <- list("/path/to/peak1.narrowPeak", "/path/to/peak2.narrowPeak")
## or SEACR peak files
peak_files <- list("/path/to/peak1.bed", "/path/to/peak2.bed")In this example, we will use the bundled GRanges
peaks:
peak_files <- list(CTCF_ChIP_peaks, CTCF_TIP_peaks)Alignment Files
Optionally provide a list of path to .bam alignment
files, or ENCODE file IDs to generate additional comparisons like FRiP
scores.
In this example, we will use the built-in alignment files.
## Alignment files
CTCF_ChIP_alignment <- system.file("extdata", "CTCF_ChIP_alignment.bam",
package = "MotifPeeker")
CTCF_TIP_alignment <- system.file("extdata", "CTCF_TIP_alignment.bam",
package = "MotifPeeker")
alignment_files <- list(CTCF_ChIP_alignment, CTCF_TIP_alignment)Motif Files
MotifPeeker accepts a list of either
universalmotif objects, or paths to the
.jaspar files.
## JASPAR motif files
motif_files <- list("/path/to/motif1.jaspar", "/path/to/motif2.jaspar")If you use JASPAR motif files, it is recommended that you label them
by using the motif_labels parameter of the
MotifPeeker() function.
In this example, we will use the bundled universalmotif
motifs:
motif_files <- list(motif_MA1102.3, motif_MA1930.2)Run MotifPeeker
The report can be generated by using the main function
MotifPeeker(). For more run customisations, refer to the
next sections.
if (MotifPeeker:::confirm_meme_install(continue = TRUE)) {
MotifPeeker(
peak_files = peak_files,
reference_index = 2, # Set TIP-seq experiment as reference
alignment_files = alignment_files,
exp_labels = c("ChIP", "TIP"),
exp_type = c("chipseq", "tipseq"),
genome_build = "hg38", # Use hg38 genome build
motif_files = motif_files,
cell_counts = NULL, # No cell-count information
motif_discovery = TRUE,
motif_discovery_count = 3, # Discover top 3 motifs
motif_db = NULL, # Use default motif database (JASPAR)
download_buttons = TRUE,
out_dir = tempdir(), # Save output in a temporary directory
BPPARAM = BiocParallel::SerialParam(), # Use two CPU cores on a 16GB RAM machine
debug = FALSE,
quiet = TRUE,
verbose = TRUE
)
}## Starting run with 1 cores.## Script run successfully.
## Output saved at: /tmp/RtmpQXcnBi/MotifPeeker_20241112_140021
## Time taken: 1.04 mins.## [1] "/tmp/RtmpQXcnBi/MotifPeeker_20241112_140021"Required Inputs
These input parameters must be provided:
Details
-
peak_files: A list of path to peak files orGRangesobjects with the peaks to analyse. Currently, only peak files fromMACS2/3(.narrowPeak) andSEACR(.bed) are supported. ENCODE file IDs can also be provided to automatically fetch peak file(s) from the ENCODE database.
-
reference_index: An integer specifying the index of the reference dataset in thepeak_fileslist to use as reference for various comparisons. (default = 1)
-
genome_build: A character string or aBSgenomeobject specifying the genome build of the datasets. At the moment, only hg38 and hg19 are supported as abbreviated input.
-
out_dir: A character string specifying the output directory to save the HTML report and other files.
Optional Inputs
These input parameters optional, but recommended to add more analyses, or enhance them:
Details
-
alignment_files: A list of path to alignment files orRsamtools::BamFileobjects with the alignment sequences to analyse. Alignment files are used to calculate read-related metrics like FRiP score. ENCODE file IDs can also be provided to automatically fetch alignment file(s) from the ENCODE database.
-
exp_labels: A character vector of labels for each peak file. If not provided, capital letters will be used as labels in the report. -
exp_type: A character vector of experimental types for each peak file.
Useful for comparison of different methods. If not provided, all datasets will be classified as “unknown” experiment types in the report.exp_typeis used only for labelling. It does not affect the analyses. You can also input custom strings. Datasets will be grouped as long as they match their respectiveexp_type. Supported experimental types are:
-chipseq: ChIP-seq data
-tipseq: TIP-seq data
-cuttag: CUT&Tag data
-cutrun: CUT&Run data
-
motif_files: A character vector of path to motif files, or a vector ofuniversalmotif-classobjects. Required to run Known Motif Enrichment Analysis. JASPAR matrix IDs can also be provided to automatically fetch motifs from the JASPAR.
-
motif_labels: A character vector of labels for each motif file. Only used if path to file names are passed in motif_files. If not provided, the motif file names will be used as labels.
-
cell_counts: An integer vector of experiment cell counts for each peak file (if available). Creates additional comparisons based on cell counts.
-
motif_db: Path to.memeformat file to use as reference database, or a list ofuniversalmotif-classobjects. Results from motif discovery are searched against this database to find similar motifs. If not provided, JASPAR CORE database will be used, making this parameter truly optional. NOTE: p-value estimates are inaccurate when the database has fewer than 50 entries.
Other Options
For more information on additional parameters, please refer to the
documentation for MotifPeeker().
Runtime Guidance
For 4 datasets, the runtime is approximately 3 minutes with motif_discovery disabled. However, motif discovery can take hours to complete.
To make computation faster, we highly recommend tuning the following arguments:
Details
-
BPPARAM = Multicore(x): Running motif discovery in parallel can significantly reduce runtime, but it is very memory-intensive, consuming upwards of 10GB of RAM per thread. Memory starvation can greatly slow the process, so set workers (x) with caution.
-
motif_discovery_count: The number of motifs to discover per sequence group exponentially increases runtime. We recommend no more than 5 motifs to make a meaningful inference.
-
trim_seq_width: Trimming sequences before running motif discovery can significantly reduce the search space. Sequence length can exponentially increase runtime. We recommend running the script withmotif_discovery = FALSEand studying the motif-summit distance distribution under general metrics to find the sequence length that captures most motifs. A good starting point is 150 but it can be reduced further if appropriate.
Outputs
MotifPeeker generates its output in a new folder within
he out_dir directory. The folder is named
MotifPeeker_YYYYMMDD_HHMMSS and contains the following
files:
-
MotifPeeker.html: The main HTML report, including all analyses and plots.
- Output from various MEME suite tools in their respecive
sub-directories, if
save_runfilesis set toTRUE.
Troubleshooting
If something does not work as expected, refer to troubleshooting.
Future Enhancements
- Add support for outputs from more peak callers.
- Automatically detect ideal
trim_peak_widthto reduce motif discovery runtime.
- Add more troubleshooting steps to the documentation.
Session Info
utils::sessionInfo()## R version 4.4.1 (2024-06-14)
## Platform: x86_64-pc-linux-gnu
## Running under: Ubuntu 22.04.4 LTS
##
## Matrix products: default
## BLAS: /usr/lib/x86_64-linux-gnu/openblas-pthread/libblas.so.3
## LAPACK: /usr/lib/x86_64-linux-gnu/openblas-pthread/libopenblasp-r0.3.20.so; LAPACK version 3.10.0
##
## locale:
## [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
## [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
## [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
## [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
## [9] LC_ADDRESS=C LC_TELEPHONE=C
## [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
##
## time zone: UTC
## tzcode source: system (glibc)
##
## attached base packages:
## [1] stats graphics grDevices utils datasets methods base
##
## other attached packages:
## [1] MotifPeeker_0.99.11
##
## loaded via a namespace (and not attached):
## [1] RColorBrewer_1.1-3 jsonlite_1.8.9
## [3] magrittr_2.0.3 farver_2.1.2
## [5] rmarkdown_2.29 fs_1.6.5
## [7] BiocIO_1.16.0 zlibbioc_1.52.0
## [9] ragg_1.3.2 vctrs_0.6.5
## [11] memoise_2.0.1 Rsamtools_2.22.0
## [13] b64_0.1.3 RCurl_1.98-1.16
## [15] webshot_0.5.5 htmltools_0.5.8.1
## [17] S4Arrays_1.6.0 curl_6.0.0
## [19] SparseArray_1.6.0 sass_0.4.9
## [21] bslib_0.8.0 htmlwidgets_1.6.4
## [23] desc_1.4.3 plyr_1.8.9
## [25] testthat_3.2.1.1 lubridate_1.9.3
## [27] plotly_4.10.4 cachem_1.1.0
## [29] GenomicAlignments_1.42.0 mime_0.12
## [31] downloadthis_0.4.1 lifecycle_1.0.4
## [33] iterators_1.0.14 pkgconfig_2.0.3
## [35] Matrix_1.7-0 R6_2.5.1
## [37] fastmap_1.2.0 GenomeInfoDbData_1.2.13
## [39] MatrixGenerics_1.18.0 digest_0.6.37
## [41] colorspace_2.1-1 ps_1.8.1
## [43] S4Vectors_0.44.0 rprojroot_2.0.4
## [45] pkgload_1.4.0 crosstalk_1.2.1
## [47] textshaping_0.4.0 GenomicRanges_1.58.0
## [49] RSQLite_2.3.7 seriation_1.5.6
## [51] bsplus_0.1.4 labeling_0.4.3
## [53] filelock_1.0.3 timechange_0.3.0
## [55] fansi_1.0.6 httr_1.4.7
## [57] abind_1.4-8 compiler_4.4.1
## [59] withr_3.0.2 bit64_4.5.2
## [61] BiocParallel_1.40.0 viridis_0.6.5
## [63] DBI_1.2.3 heatmaply_1.5.0
## [65] dendextend_1.18.1 R.utils_2.12.3
## [67] MASS_7.3-61 DelayedArray_0.32.0
## [69] rjson_0.2.23 tools_4.4.1
## [71] zip_2.3.1 ggseqlogo_0.2
## [73] R.oo_1.27.0 glue_1.8.0
## [75] restfulr_0.0.15 grid_4.4.1
## [77] reshape2_1.4.4 generics_0.1.3
## [79] gtable_0.3.6 BSgenome_1.74.0
## [81] tzdb_0.4.0 R.methodsS3_1.8.2
## [83] ca_0.71.1 tidyr_1.3.1
## [85] data.table_1.16.2 hms_1.1.3
## [87] xml2_1.3.6 utf8_1.2.4
## [89] XVector_0.46.0 BiocGenerics_0.52.0
## [91] memes_1.14.0 stringr_1.5.1
## [93] foreach_1.5.2 pillar_1.9.0
## [95] vroom_1.6.5 dplyr_1.1.4
## [97] BiocFileCache_2.14.0 lattice_0.22-6
## [99] rtracklayer_1.66.0 bit_4.5.0
## [101] universalmotif_1.24.0 tidyselect_1.2.1
## [103] BSgenome.Hsapiens.UCSC.hg38_1.4.5 registry_0.5-1
## [105] Biostrings_2.74.0 knitr_1.49
## [107] gridExtra_2.3 IRanges_2.40.0
## [109] SummarizedExperiment_1.36.0 cmdfun_1.0.2
## [111] stats4_4.4.1 xfun_0.49
## [113] Biobase_2.66.0 brio_1.1.5
## [115] matrixStats_1.4.1 DT_0.33
## [117] stringi_1.8.4 UCSC.utils_1.2.0
## [119] lazyeval_0.2.2 yaml_2.3.10
## [121] evaluate_1.0.1 codetools_0.2-20
## [123] tibble_3.2.1 cli_3.6.3
## [125] systemfonts_1.1.0 processx_3.8.4
## [127] munsell_0.5.1 jquerylib_0.1.4
## [129] Rcpp_1.0.13-1 GenomeInfoDb_1.42.0
## [131] emoji_16.0.0 dbplyr_2.5.0
## [133] XML_3.99-0.17 parallel_4.4.1
## [135] pkgdown_2.1.1 ggplot2_3.5.1
## [137] readr_2.1.5 assertthat_0.2.1
## [139] blob_1.2.4 bitops_1.0-9
## [141] viridisLite_0.4.2 scales_1.3.0
## [143] purrr_1.0.2 crayon_1.5.3
## [145] rlang_1.1.4 TSP_1.2-4
## [147] waldo_0.6.1