Contents

1 Introduction

This package is designed to enable the Gene Regulatory Analysis using Variable IP (GRAVI) workflow, as a method for detecting differential binding in ChIP-Seq datasets. As a workflow focussed on data integration, most functions provided by the package extraChIPs are designed to enable comparison across datasets. This vignette looks primarily at functions which work with GenomicRanges objects.

2 Installation

In order to use the package extraChIPs and follow this vignette, we recommend using the package BiocManager hosted on CRAN. Once this is installed, the additional packages required for this vignette (tidyverse, plyranges and Gviz) can also be installed.

if (!"BiocManager" %in% rownames(installed.packages()))
  install.packages("BiocManager")
BiocManager::install(c("tidyverse", "plyranges", "Gviz"))
BiocManager::install("extraChIPs")

3 Coercion

The advent of the tidyverse has led to tibble objects becoming a common alternative to data.frame or DataFrame objects. Simple functions within extraChIP enable coercion from GRanges, GInteractions and DataFrame objects to tibble objects, with list columns correctly handled. By default these coercion functions will coerce GRanges elements to a character vector. Similarly, GRanges represented as a character column can be coerced to the ranges element of a GRanges object.

First let’s coerce from a tibble (or S3 data.frame) to a GRanges

library(tidyverse)
library(extraChIPs)
set.seed(73)
df <- tibble(
  range = c("chr1:1-10:+", "chr1:5-10:+", "chr1:5-6:+"),
  gene_id = "gene1",
  tx_id = paste0("transcript", 1:3),
  score = runif(3)
)
df
## # A tibble: 3 × 4
##   range       gene_id tx_id        score
##   <chr>       <chr>   <chr>        <dbl>
## 1 chr1:1-10:+ gene1   transcript1 0.442 
## 2 chr1:5-10:+ gene1   transcript2 0.0831
## 3 chr1:5-6:+  gene1   transcript3 0.615
gr <- colToRanges(df, "range")
gr
## GRanges object with 3 ranges and 3 metadata columns:
##       seqnames    ranges strand |     gene_id       tx_id     score
##          <Rle> <IRanges>  <Rle> | <character> <character> <numeric>
##   [1]     chr1      1-10      + |       gene1 transcript1 0.4423369
##   [2]     chr1      5-10      + |       gene1 transcript2 0.0831099
##   [3]     chr1       5-6      + |       gene1 transcript3 0.6146112
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

Coercion back to a tibble will place the ranges as a character column by default. However, this can be turned off and the conventional coercion from as.data.frame will instead be applied, internally wrapped in as_tibble()

as_tibble(gr)
## # A tibble: 3 × 4
##   range       gene_id tx_id        score
##   <chr>       <chr>   <chr>        <dbl>
## 1 chr1:1-10:+ gene1   transcript1 0.442 
## 2 chr1:5-10:+ gene1   transcript2 0.0831
## 3 chr1:5-6:+  gene1   transcript3 0.615
as_tibble(gr, rangeAsChar = FALSE)
## # A tibble: 3 × 8
##   seqnames start   end width strand gene_id tx_id        score
##   <fct>    <int> <int> <int> <fct>  <chr>   <chr>        <dbl>
## 1 chr1         1    10    10 +      gene1   transcript1 0.442 
## 2 chr1         5    10     6 +      gene1   transcript2 0.0831
## 3 chr1         5     6     2 +      gene1   transcript3 0.615

A simple feature which may be useful for printing gene names using rmarkdown is contained in collapseGenes(). Here a character vector of gene names is collapsed into a glue object of length `, with gene names rendered in italics by default.

gn <- c("Gene1", "Gene2", "Gene3")
collapseGenes(gn)

Gene1, Gene2 and Gene3

4 Formation of Consensus Peaks

The formation of consensus peaks incorporating ranges from multiple replicates is a key part of many ChIP-Seq analyses. A common format returned by tools such as mcas2 callpeak is the narrowPeak (or broadPeak) format. Sets of these across multiple replicates can imported using the function importPeaks(), which returns a GRangesList()

fl <- system.file(
    c("extdata/ER_1.narrowPeak", "extdata/ER_2.narrowPeak"),
    package = "extraChIPs"
)
peaks <- importPeaks(fl)
names(peaks) <- c("ER_1", "ER_2")

In the above we have loaded the peaks from two replicates from the Estrogen Receptor (ER) in T-47D cells. To form consensus peaks we can place a restriction on the number of replicates an overlapping peak needs to appear in. By default, the function makeConsensus() sets the proportion to be zero (p = 0) so all peaks are retained. Note that a logical vector/column is returned for each replicate, along with the number of replicates the consensus peak is derived from.

makeConsensus(peaks)
## GRanges object with 14 ranges and 3 metadata columns:
##        seqnames          ranges strand |      ER_1      ER_2         n
##           <Rle>       <IRanges>  <Rle> | <logical> <logical> <numeric>
##    [1]     chr1   856458-856640      * |      TRUE     FALSE         1
##    [2]     chr1   868541-868839      * |     FALSE      TRUE         1
##    [3]     chr1 1008550-1010075      * |      TRUE      TRUE         2
##    [4]     chr1 1014770-1016015      * |      TRUE      TRUE         2
##    [5]     chr1 1051307-1051918      * |      TRUE      TRUE         2
##    ...      ...             ...    ... .       ...       ...       ...
##   [10]     chr1 1608098-1608346      * |      TRUE      TRUE         2
##   [11]     chr1 1690460-1690641      * |     FALSE      TRUE         1
##   [12]     chr1 1790733-1790975      * |      TRUE     FALSE         1
##   [13]     chr1 1878927-1879257      * |      TRUE      TRUE         2
##   [14]     chr1 1900588-1900902      * |      TRUE     FALSE         1
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

However, we may wish for peaks to appear in all replicates, so we can set the argument p = 1 (or any other value \(0 \leq p \leq 1\)).

makeConsensus(peaks, p = 1)
## GRanges object with 6 ranges and 3 metadata columns:
##       seqnames          ranges strand |      ER_1      ER_2         n
##          <Rle>       <IRanges>  <Rle> | <logical> <logical> <numeric>
##   [1]     chr1 1008550-1010075      * |      TRUE      TRUE         2
##   [2]     chr1 1014770-1016015      * |      TRUE      TRUE         2
##   [3]     chr1 1051307-1051918      * |      TRUE      TRUE         2
##   [4]     chr1 1368372-1369009      * |      TRUE      TRUE         2
##   [5]     chr1 1608098-1608346      * |      TRUE      TRUE         2
##   [6]     chr1 1878927-1879257      * |      TRUE      TRUE         2
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

In addition, we may wish to keep the values from the mcols as part of our consensus peaks, such as the qValue. This can be specified using the var argument, and will return a CompressedList as returned by reduceMC(), seen below.

makeConsensus(peaks, p = 1, var = "qValue")
## GRanges object with 6 ranges and 4 metadata columns:
##       seqnames          ranges strand |          qValue      ER_1      ER_2
##          <Rle>       <IRanges>  <Rle> |   <NumericList> <logical> <logical>
##   [1]     chr1 1008550-1010075      * | 635.748,605.040      TRUE      TRUE
##   [2]     chr1 1014770-1016015      * | 95.1223,78.3953      TRUE      TRUE
##   [3]     chr1 1051307-1051918      * | 65.5538,95.9677      TRUE      TRUE
##   [4]     chr1 1368372-1369009      * | 22.5483,29.1231      TRUE      TRUE
##   [5]     chr1 1608098-1608346      * | 72.3235,64.8439      TRUE      TRUE
##   [6]     chr1 1878927-1879257      * | 33.8900,14.4909      TRUE      TRUE
##               n
##       <numeric>
##   [1]         2
##   [2]         2
##   [3]         2
##   [4]         2
##   [5]         2
##   [6]         2
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

We could even identify the peak centre and pass these to the set of consensus peaks for downstream peak re-centreing.

library(plyranges)
peaks %>% 
    endoapply(mutate, centre = start + peak) %>% 
    makeConsensus(p = 1, var = "centre")
## GRanges object with 6 ranges and 4 metadata columns:
##       seqnames          ranges strand |          centre      ER_1      ER_2
##          <Rle>       <IRanges>  <Rle> |   <NumericList> <logical> <logical>
##   [1]     chr1 1008550-1010075      * | 1009212,1009212      TRUE      TRUE
##   [2]     chr1 1014770-1016015      * | 1014981,1014949      TRUE      TRUE
##   [3]     chr1 1051307-1051918      * | 1051565,1051634      TRUE      TRUE
##   [4]     chr1 1368372-1369009      * | 1368613,1368767      TRUE      TRUE
##   [5]     chr1 1608098-1608346      * | 1608247,1608247      TRUE      TRUE
##   [6]     chr1 1878927-1879257      * | 1879087,1879094      TRUE      TRUE
##               n
##       <numeric>
##   [1]         2
##   [2]         2
##   [3]         2
##   [4]         2
##   [5]         2
##   [6]         2
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

We could then find the mean of the peak centres for an averaged centre.

peaks %>% 
    endoapply(mutate, centre = start + peak) %>% 
    makeConsensus(p = 1, var = "centre") %>% 
    mutate(centre = vapply(centre, mean, numeric(1)))
## GRanges object with 6 ranges and 4 metadata columns:
##       seqnames          ranges strand |    centre      ER_1      ER_2         n
##          <Rle>       <IRanges>  <Rle> | <numeric> <logical> <logical> <numeric>
##   [1]     chr1 1008550-1010075      * |   1009212      TRUE      TRUE         2
##   [2]     chr1 1014770-1016015      * |   1014965      TRUE      TRUE         2
##   [3]     chr1 1051307-1051918      * |   1051600      TRUE      TRUE         2
##   [4]     chr1 1368372-1369009      * |   1368690      TRUE      TRUE         2
##   [5]     chr1 1608098-1608346      * |   1608247      TRUE      TRUE         2
##   [6]     chr1 1878927-1879257      * |   1879090      TRUE      TRUE         2
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

5 Simple Operations Retaining mcols()

The standard set operations implemented in the package GenomicRanges will always drop the mcols element by default. The extraChIPs functions reduceMC(), setdiffMC(), intersectMC() and unionMC() all produce the same output as their similarly-named functions, however, the mcols() elements in the query object are also returned. Where required, columns are coerced into CompressedList columns. This can particularly useful when needed to propagate the information contained in the initial ranges through to subsequent analytic steps

5.1 Simplifying single GRanges objects

The classical approach to defining TSS regions for a set of transcripts would be to use the function resize()`, setting the width as 1.

tss <- resize(gr, width = 1)
tss
## GRanges object with 3 ranges and 3 metadata columns:
##       seqnames    ranges strand |     gene_id       tx_id     score
##          <Rle> <IRanges>  <Rle> | <character> <character> <numeric>
##   [1]     chr1         1      + |       gene1 transcript1 0.4423369
##   [2]     chr1         5      + |       gene1 transcript2 0.0831099
##   [3]     chr1         5      + |       gene1 transcript3 0.6146112
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

As we can see, two transcripts start at position 5, so we may choose to reduce this, which would lose the mcols element. The alternative reduceMC() will retain all mcols.

GenomicRanges::reduce(tss)
## GRanges object with 2 ranges and 0 metadata columns:
##       seqnames    ranges strand
##          <Rle> <IRanges>  <Rle>
##   [1]     chr1         1      +
##   [2]     chr1         5      +
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths
reduceMC(tss)
## GRanges object with 2 ranges and 3 metadata columns:
##       seqnames    ranges strand |     gene_id                   tx_id
##          <Rle> <IRanges>  <Rle> | <character>         <CharacterList>
##   [1]     chr1         1      + |       gene1             transcript1
##   [2]     chr1         5      + |       gene1 transcript2,transcript3
##                     score
##             <NumericList>
##   [1]            0.442337
##   [2] 0.0831099,0.6146112
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

By default, this function will attempt to coerce mcols to a new mcol of the appropriate type, however, when multiple values are inevitable such as for the tx_id column above, these will be coerced to a CompressedList. The simplification of the multiple values seen in the gene_id can also be turned off if desired should repeated values be important for downstream analysis.

reduceMC(tss, simplify = FALSE) 
## GRanges object with 2 ranges and 3 metadata columns:
##       seqnames    ranges strand |         gene_id                   tx_id
##          <Rle> <IRanges>  <Rle> | <CharacterList>         <CharacterList>
##   [1]     chr1         1      + |           gene1             transcript1
##   [2]     chr1         5      + |     gene1,gene1 transcript2,transcript3
##                     score
##             <NumericList>
##   [1]            0.442337
##   [2] 0.0831099,0.6146112
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

This allows for simple integration with tidyverse nesting strategies.

reduceMC(tss, simplify = FALSE) %>% 
  as_tibble() %>% 
  unnest(everything())
## # A tibble: 3 × 4
##   range    gene_id tx_id        score
##   <chr>    <chr>   <chr>        <dbl>
## 1 chr1:1:+ gene1   transcript1 0.442 
## 2 chr1:5:+ gene1   transcript2 0.0831
## 3 chr1:5:+ gene1   transcript3 0.615

Whilst reduceMC relies on the range-reduction as implemented in GenomicRanges::reduce(), some alternative approaches are included, such as chopMC(), which finds identical ranges and nests the mcols element as CompressedList objects.

chopMC(tss)
## GRanges object with 2 ranges and 3 metadata columns:
##       seqnames    ranges strand |     gene_id                   tx_id
##          <Rle> <IRanges>  <Rle> | <character>         <CharacterList>
##   [1]     chr1         1      + |       gene1             transcript1
##   [2]     chr1         5      + |       gene1 transcript2,transcript3
##                     score
##             <NumericList>
##   [1]            0.442337
##   [2] 0.0831099,0.6146112
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

In the case of the object tss, this output is identical to reduceMC(), however, given that there are no identical ranges in gr the two functions would behave very differently on that object.

A final operation for simplifying GRanges objects would be distinctMC() which is a wrapper to dplyr]::distinct incorporating both the range and mcols. The columns to search can be called using <data-masking> approaches as detailed in the manual.

distinctMC(tss)
## GRanges object with 2 ranges and 0 metadata columns:
##       seqnames    ranges strand
##          <Rle> <IRanges>  <Rle>
##   [1]     chr1         1      +
##   [2]     chr1         5      +
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths
distinctMC(tss, gene_id)
## GRanges object with 2 ranges and 1 metadata column:
##       seqnames    ranges strand |     gene_id
##          <Rle> <IRanges>  <Rle> | <character>
##   [1]     chr1         1      + |       gene1
##   [2]     chr1         5      + |       gene1
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

5.2 Set Operations with Two GRanges Objects

Whilst reduce/reduceMC is applied to a single GRanges object, the set operation functions intersect, setdiff and union are valuable approaches for comparing ranges. Using the *MC() functions will retain mcols elements from the query range.

peaks <- GRanges(c("chr1:1-5", "chr1:9-12:*"))
peaks$peak_id <- c("peak1", "peak2")
GenomicRanges::intersect(gr, peaks, ignore.strand = TRUE)
## GRanges object with 2 ranges and 0 metadata columns:
##       seqnames    ranges strand
##          <Rle> <IRanges>  <Rle>
##   [1]     chr1       1-5      *
##   [2]     chr1      9-10      *
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths
intersectMC(gr, peaks, ignore.strand = TRUE)
## GRanges object with 2 ranges and 3 metadata columns:
##       seqnames    ranges strand |     gene_id
##          <Rle> <IRanges>  <Rle> | <character>
##   [1]     chr1       1-5      * |       gene1
##   [2]     chr1      9-10      * |       gene1
##                                     tx_id                         score
##                           <CharacterList>                 <NumericList>
##   [1] transcript1,transcript2,transcript3 0.4423369,0.0831099,0.6146112
##   [2]             transcript1,transcript2           0.4423369,0.0831099
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths
setdiffMC(gr, peaks, ignore.strand = TRUE)
## GRanges object with 1 range and 3 metadata columns:
##       seqnames    ranges strand |     gene_id
##          <Rle> <IRanges>  <Rle> | <character>
##   [1]     chr1       6-8      * |       gene1
##                                     tx_id                         score
##                           <CharacterList>                 <NumericList>
##   [1] transcript1,transcript2,transcript3 0.4423369,0.0831099,0.6146112
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths
unionMC(gr, peaks, ignore.strand = TRUE)
## GRanges object with 1 range and 3 metadata columns:
##       seqnames    ranges strand |     gene_id
##          <Rle> <IRanges>  <Rle> | <character>
##   [1]     chr1      1-12      * |       gene1
##                                     tx_id                         score
##                           <CharacterList>                 <NumericList>
##   [1] transcript1,transcript2,transcript3 0.4423369,0.0831099,0.6146112
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

There is a performance overhead to preparation of mcols as CompressedList objects and all mcols in the query object will be returned. If only wishing to retain a subset of mcols, these should be selected prior to passing to these functions.

gr %>% 
  select(tx_id) %>% 
  intersectMC(peaks, ignore.strand = TRUE)
## GRanges object with 2 ranges and 1 metadata column:
##       seqnames    ranges strand |                               tx_id
##          <Rle> <IRanges>  <Rle> |                     <CharacterList>
##   [1]     chr1       1-5      * | transcript1,transcript2,transcript3
##   [2]     chr1      9-10      * |             transcript1,transcript2
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

5.3 Overlapping proportions

Whilst the functions findOverlaps() and overlapsAny() are extremely useful, the addition of propOverlap() returns a numeric vector with the proportion of each query range (x) which overlaps any range in the subject range (y)

propOverlap(gr, peaks)
## [1] 0.7 0.5 0.5

This is also extended to enable comparison across multiple features and to classify each peak by which features that it overlaps the most strongly.

bestOverlap(gr, peaks, var = "peak_id")
## [1] "peak1" "peak2" "peak1"

6 More Complex Operations

6.1 Range Partitioning

In addition to standard set operations, one set of ranges can be used to partition another set of ranges, returning mcols from both ranges. Ranges from the query range (x) are returned after being partitioned by the ranges in the subject range (y). Subject ranges used for partitioning must be non-overlapping, and if overlapping ranges are provided, these will be reduced prior to partitioning.

This enables the identification of specific ranges from the query range (x) which overlap ranges from the subject range (y) Under this approach, mcols from both query and subject ranges will be returned to enable the clear ranges which are common and distinct within the two sets of ranges.

partitionRanges(gr, peaks)
## GRanges object with 5 ranges and 4 metadata columns:
##       seqnames    ranges strand |     peak_id     gene_id
##          <Rle> <IRanges>  <Rle> | <character> <character>
##   [1]     chr1       1-5      + |       peak1       gene1
##   [2]     chr1         5      + |       peak1       gene1
##   [3]     chr1         6      + |        <NA>       gene1
##   [4]     chr1       6-8      + |        <NA>       gene1
##   [5]     chr1      9-10      + |       peak2       gene1
##                         tx_id               score
##               <CharacterList>       <NumericList>
##   [1]             transcript1            0.442337
##   [2] transcript2,transcript3 0.0831099,0.6146112
##   [3]             transcript3            0.614611
##   [4] transcript1,transcript2 0.4423369,0.0831099
##   [5] transcript1,transcript2 0.4423369,0.0831099
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

Whilst this shares some similarity with intersectMC() the additional capabilities provide greater flexibility.

partitionRanges(gr, peaks) %>% 
  subset(is.na(peak_id))
## GRanges object with 2 ranges and 4 metadata columns:
##       seqnames    ranges strand |     peak_id     gene_id
##          <Rle> <IRanges>  <Rle> | <character> <character>
##   [1]     chr1         6      + |        <NA>       gene1
##   [2]     chr1       6-8      + |        <NA>       gene1
##                         tx_id               score
##               <CharacterList>       <NumericList>
##   [1]             transcript3            0.614611
##   [2] transcript1,transcript2 0.4423369,0.0831099
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

6.2 Stitching Ranges

Using the function stitchRanges() we are able to join together sets of nearby ranges, but with the option of placing clear barriers between ranges, across which ranges cannot be joined. This may be useful if joining enhancers to form putative super-enhancers, but explicitly omitting defined promoter regions.

enh <- GRanges(c("chr1:1-10", "chr1:101-110", "chr1:181-200"))
prom <- GRanges("chr1:150:+")
se <- stitchRanges(enh, exclude = prom, maxgap = 100)
se
## GRanges object with 2 ranges and 0 metadata columns:
##       seqnames    ranges strand
##          <Rle> <IRanges>  <Rle>
##   [1]     chr1     1-110      *
##   [2]     chr1   181-200      *
##   -------
##   seqinfo: 1 sequence from an unspecified genome; no seqlengths

As a visualisation (below) ranges within 100bp were stitched together, however the region defined as a ‘promoter’ acted as a barrier and ranges were not stitched together across this barrier.

7 Session Info

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:   /home/biocbuild/bbs-3.19-bioc/R/lib/libRblas.so 
## LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.10.0
## 
## locale:
##  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
##  [3] LC_TIME=en_GB              LC_COLLATE=C              
##  [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: America/New_York
## tzcode source: system (glibc)
## 
## attached base packages:
## [1] stats4    stats     graphics  grDevices utils     datasets  methods  
## [8] base     
## 
## other attached packages:
##  [1] magrittr_2.0.3              quantro_1.38.0             
##  [3] here_1.0.1                  ggrepel_0.9.5              
##  [5] glue_1.7.0                  scales_1.3.0               
##  [7] plyranges_1.24.0            extraChIPs_1.8.5           
##  [9] ggside_0.3.1                patchwork_1.2.0            
## [11] edgeR_4.2.1                 limma_3.60.4               
## [13] rtracklayer_1.64.0          BiocParallel_1.38.0        
## [15] csaw_1.38.0                 SummarizedExperiment_1.34.0
## [17] Biobase_2.64.0              MatrixGenerics_1.16.0      
## [19] matrixStats_1.3.0           Rsamtools_2.20.0           
## [21] Biostrings_2.72.1           XVector_0.44.0             
## [23] GenomicRanges_1.56.1        GenomeInfoDb_1.40.1        
## [25] IRanges_2.38.1              S4Vectors_0.42.1           
## [27] BiocGenerics_0.50.0         lubridate_1.9.3            
## [29] forcats_1.0.0               stringr_1.5.1              
## [31] dplyr_1.1.4                 purrr_1.0.2                
## [33] readr_2.1.5                 tidyr_1.3.1                
## [35] tibble_3.2.1                ggplot2_3.5.1              
## [37] tidyverse_2.0.0             BiocStyle_2.32.1           
## 
## loaded via a namespace (and not attached):
##   [1] ProtGenerics_1.36.0        bitops_1.0-7              
##   [3] httr_1.4.7                 RColorBrewer_1.1-3        
##   [5] doParallel_1.0.17          InteractionSet_1.32.0     
##   [7] tools_4.4.1                doRNG_1.8.6               
##   [9] backports_1.5.0            utf8_1.2.4                
##  [11] R6_2.5.1                   HDF5Array_1.32.0          
##  [13] mgcv_1.9-1                 lazyeval_0.2.2            
##  [15] Gviz_1.48.0                rhdf5filters_1.16.0       
##  [17] withr_3.0.0                prettyunits_1.2.0         
##  [19] gridExtra_2.3              base64_2.0.1              
##  [21] VennDiagram_1.7.3          preprocessCore_1.66.0     
##  [23] cli_3.6.3                  formatR_1.14              
##  [25] labeling_0.4.3             sass_0.4.9                
##  [27] genefilter_1.86.0          askpass_1.2.0             
##  [29] foreign_0.8-87             siggenes_1.78.0           
##  [31] illuminaio_0.46.0          dichromat_2.0-0.1         
##  [33] scrime_1.3.5               BSgenome_1.72.0           
##  [35] rstudioapi_0.16.0          RSQLite_2.3.7             
##  [37] generics_0.1.3             BiocIO_1.14.0             
##  [39] Matrix_1.7-0               interp_1.1-6              
##  [41] futile.logger_1.4.3        fansi_1.0.6               
##  [43] abind_1.4-5                lifecycle_1.0.4           
##  [45] yaml_2.3.10                rhdf5_2.48.0              
##  [47] SparseArray_1.4.8          BiocFileCache_2.12.0      
##  [49] grid_4.4.1                 blob_1.2.4                
##  [51] crayon_1.5.3               lattice_0.22-6            
##  [53] ComplexUpset_1.3.3         GenomicFeatures_1.56.0    
##  [55] annotate_1.82.0            KEGGREST_1.44.1           
##  [57] pillar_1.9.0               knitr_1.48                
##  [59] beanplot_1.3.1             metapod_1.12.0            
##  [61] rjson_0.2.21               codetools_0.2-20          
##  [63] data.table_1.15.4          vctrs_0.6.5               
##  [65] png_0.1-8                  gtable_0.3.5              
##  [67] cachem_1.1.0               xfun_0.46                 
##  [69] S4Arrays_1.4.1             survival_3.7-0            
##  [71] iterators_1.0.14           statmod_1.5.0             
##  [73] GenomicInteractions_1.38.0 nlme_3.1-165              
##  [75] bit64_4.0.5                progress_1.2.3            
##  [77] filelock_1.0.3             rprojroot_2.0.4           
##  [79] bslib_0.7.0                nor1mix_1.3-3             
##  [81] rpart_4.1.23               colorspace_2.1-1          
##  [83] DBI_1.2.3                  Hmisc_5.1-3               
##  [85] nnet_7.3-19                tidyselect_1.2.1          
##  [87] bit_4.0.5                  compiler_4.4.1            
##  [89] curl_5.2.1                 httr2_1.0.2               
##  [91] htmlTable_2.4.3            xml2_1.3.6                
##  [93] DelayedArray_0.30.1        bookdown_0.40             
##  [95] checkmate_2.3.1            quadprog_1.5-8            
##  [97] rappdirs_0.3.3             digest_0.6.36             
##  [99] rmarkdown_2.27             GEOquery_2.72.0           
## [101] htmltools_0.5.8.1          pkgconfig_2.0.3           
## [103] jpeg_0.1-10                base64enc_0.1-3           
## [105] sparseMatrixStats_1.16.0   highr_0.11                
## [107] dbplyr_2.5.0               fastmap_1.2.0             
## [109] ensembldb_2.28.0           rlang_1.1.4               
## [111] htmlwidgets_1.6.4          UCSC.utils_1.0.0          
## [113] DelayedMatrixStats_1.26.0  farver_2.1.2              
## [115] jquerylib_0.1.4            jsonlite_1.8.8            
## [117] mclust_6.1.1               VariantAnnotation_1.50.0  
## [119] RCurl_1.98-1.16            Formula_1.2-5             
## [121] GenomeInfoDbData_1.2.12    Rhdf5lib_1.26.0           
## [123] munsell_0.5.1              Rcpp_1.0.13               
## [125] stringi_1.8.4              zlibbioc_1.50.0           
## [127] MASS_7.3-61                plyr_1.8.9                
## [129] bumphunter_1.46.0          minfi_1.50.0              
## [131] parallel_4.4.1             deldir_2.0-4              
## [133] splines_4.4.1              multtest_2.60.0           
## [135] hms_1.1.3                  locfit_1.5-9.10           
## [137] igraph_2.0.3               rngtools_1.5.2            
## [139] biomaRt_2.60.1             futile.options_1.0.1      
## [141] XML_3.99-0.17              evaluate_0.24.0           
## [143] latticeExtra_0.6-30        biovizBase_1.52.0         
## [145] lambda.r_1.2.4             BiocManager_1.30.23       
## [147] tzdb_0.4.0                 foreach_1.5.2             
## [149] tweenr_2.0.3               openssl_2.2.0             
## [151] polyclip_1.10-7            reshape_0.8.9             
## [153] ggforce_0.4.2              broom_1.0.6               
## [155] xtable_1.8-4               restfulr_0.0.15           
## [157] AnnotationFilter_1.28.0    memoise_2.0.1             
## [159] AnnotationDbi_1.66.0       GenomicAlignments_1.40.0  
## [161] cluster_2.1.6              timechange_0.3.0