NEWS
epiSeeker 1.1.4
annotateSeq() keeps annotation, geneChr/geneStart/geneEnd, geneId and
transcriptId consistent with each other at transcript level (issue #252 of
ChIPseeker). The transcript id of an exon/intron/UTR hit was taken from
names(genomicRegion)[subjectIndex], indexing the unlisted ranges with the
names of the GRangesList; this mostly returned NA and occasionally a wrong
transcript, so a peak could be reported with the metadata of an unrelated
transcript (even on another chromosome). The ids are now expanded before
indexing, and the same alignment is applied at level = "gene", where the gene
of the overlapping exon/intron/UTR is resolved with TXID2EG() and used
instead of the nearest gene. (2026-10-01, Thu)seq2gene() no longer fails with $ operator is invalid for atomic vectors
when none of the queried regions overlaps an exon
(getGenomicAnnotation.internal() returns NA then); host-gene extraction is
skipped and the nearest/flanking genes are still reported (issue #248 of
ChIPseeker). (2026-10-01, Thu)downloadGEObedFiles()/downloadGSMbedFiles() rewrite the ftp:// urls of
gsminfo$supplementary_file to https:// before downloading and report the
underlying download error when a file cannot be fetched (issue #254 of
ChIPseeker). (2026-10-01, Thu)- Documented the distance definition of
flank_gene_distances reported by
annotateSeq(..., addFlankGeneInfo = TRUE): a distance of 0 means that the
peak overlaps the feature range - at level = "transcript" the feature is the
whole transcript, which is why most entries can be 0 - while non-overlapping
peaks get the signed distance to the feature TSS (issue #235 of ChIPseeker).
(2026-10-01, Thu)
annotateSeq() no longer fails with Error: invalid subscript when every
peak is dropped, which typically happens because none of the seqlevels of the
peaks matches TxDb (chr1 vs NC_000001.11): .get_distance_to_gene_end()
no longer hands the SortedByQueryHits object returned by follow() for an
empty query to features[], and annotateSeq() reports the seqlevels mismatch
together with how to align the chromosome names (seqlevelsStyle()).
(2026-10-01, Thu)enrichPeakOverlap()/enrichAnnoOverlap() accept a single GRanges as
targetPeak: it was passed on unwrapped while the overlap code works on a
list of target peak sets, so the call failed with "GRanges objects don't
support [[, as.list(), lapply()". The documentation of enrichPeakOverlap()
and enrichAnnoOverlap() now also states the direction of the test (the
observed ratio is the fraction of target peaks covered by the query peaks and
the target is the shuffled set, so N_OL is direction free while the p-value
is not) and that N_OL of enrichAnnoOverlap() counts genes and can therefore
exceed the number of input peaks (issue #84 of ChIPseeker).
(2026-10-01, Thu)- An empty input is now handled:
annotateSeq(GRanges(), ...) returns an empty
csAnno instead of failing with "Error: invalid subscript". The root cause
(follow()/precede() returning a SortedByQueryHits object instead of indices for
an empty query) is guarded in getNearestFeatureIndicesAndDistances(), which
also makes seq2gene() return character(0) for an empty input.
(2026-10-01, Thu)
enrichPeakOverlap() gained an opt-in symmetric argument. With
symmetric = TRUE the mirrored direction (query and target exchanged) is
computed as well and the two one-sided permutation p-values are combined as
min(1, 2*min(p, p_rev)) (Hedges), so the result no longer depends on the order
of the arguments. The default stays FALSE (one-sided) and all existing numbers
are unchanged; the mirrored test doubles the number of permutations and raises
the smallest reportable p-value to 2/(nShuffle+1). enrichAnnoOverlap() needs
no such argument, its hypergeometric p-value is already exactly symmetric (issue
#84 of ChIPseeker). (2026-10-01, Thu)upsetplot(vennpie = TRUE) can be drawn again with ggplot2 >= 4.0. The vennpie
sub-view is embedded as an annotation_custom() layer, which ggplot2 only
supports below coord_cartesian(). The sub-view no longer uses coord_fixed()
and the parent plot is converted with ggplotify as well, because the ggupset
coordinate system (CoordCombMatrix) rejects such layers. No replacement for
the removed coord_fixed() is added: ggplotify rasterises the grob with the
aspect ratio of the device, and forcing a square panel distorted the venn
diagram further (measured anisotropy 1.43 without versus 1.50 with
theme(aspect.ratio = 1), against 1.41 for the undistorted drawing). Before,
the plot failed with `annotation_custom()` only works with `coord_cartesian()`. Note that ggupset 0.4.1 itself still maps the
deprecated size aesthetic, so drawing the plot may print
"Please use linewidth instead"; the warning comes from ggupset
(https://github.com/const-ae/ggupset/issues) and can be ignored.
(2026-10-01, Thu)
epiSeeker 1.1.3
annotateSeq() now reports geneChr and geneStrand as characters instead of
factor codes. as.data.frame() returns 'seqnames'/'strand' as factors and
assigning a factor into mcols() dropped the class, so both columns came out
as integers (e.g. geneStrand = 1/2 instead of +/-, and a wrong geneChr
whenever the seqlevels were not in numeric order). (2026-09-15, Tue)annotateSeq() now warns when peaks are dropped because no feature of TxDb
can be found for them (e.g. peaks on contigs/scaffolds without genes, or a
seqlevels style mismatch). Previously they disappeared silently. (2026-09-15, Tue)annotateSeq(..., sameStrand = TRUE) no longer assigns a peak to a feature on
the opposite strand. Overlap detection in getNearestFeatureIndicesAndDistances()
was calling findOverlaps() with unstrand(features), which bypassed sameStrand
and overrode the strand-aware nearest-feature result. Peaks with ambiguous
strand (*) are unaffected and still match features on any strand.
(2026-09-15, Tue)plotAnnoBar() no longer uses the deprecated ggplot2::aes_string(). It follows
the tidy evaluation idiom already used by plotDistToTSS(). (2026-09-15, Tue)- dropped unused
@importFrom directives with no call site (aplot::xlim2,
ggplot2::geom_segment, ggplot2::geom_text, ggplot2::scale_fill_hue,
utils::getFromNamespace) and a duplicate ggplot2::scale_fill_brewer. The
grid::unit() import, previously declared in upsetplot() which never used it,
now sits in plotBmProf() where unit() is actually called, keeping grid a
genuine import. (2026-09-17, Thu)
epiSeeker 1.1.2
- documentation build migrated to roxygen2 8.1.0 (re-generated Rd pages and
NAMESPACE; fixed
csAnno class @aliases to be a single line); the exported
API is unchanged
- fixed bug in
getNearestFeatureIndicesAndDistances() where results for
overlap=="all" were silently overridden by the overlap=="TSS" branch,
causing the two modes to behave identically (2026-09-07, Mon)
epiSeeker 1.1.1
- make csAnno subset robust to GRanges metadata columns (2027-07-07, Tue)
epiSeeker 1.0.0
- Bioconductor RELEASE_3_23 (2026-04-29, Wed)
epiSeeker 0.99.12
- add more runnable examples and increase the coverage of tests (44.18%) (2025-11-30, Sun)
- use new demo data for base modification (
demo_bmdata) and move some dependency packages from 'Imports' to 'Suggests' (2025-11-10, Mon)
- consistently use 'hg38' for demo with new demo file, a small subset derived from GSM6418464 (2025-11-07, Fri)
- fixed Notes reported by
BiocCheck and add 'GeneRegulation' to 'biocViews' (2025-11-07, Fri)
- update test for the changes of TxDb.Hsapiens.UCSC.hg19.knownGene (2025-11-06, Thu)
- new cache mechanism from 'yulab.utils' (2025-10-15, Wed)
- fixed R check (2025-10-05, Sun)
- remove dependent of
Vennerable package (2025-09-27, Sat)
- epiSeeker inherits from ChIPseeker to support analysis of multi-omics epigenomic data (2025-09-19, Fri)
- epiSeeker inherits functions from ChIPseeker to plot peak profile, plot coverage, peak annotaions, visualization of annotation, data mining and statistic for peak overla
- epiSeeker merges multiple functions to plot profile (getTagMatrix + plotPeakProf/plotPeakHeatmap), and add featrues of dealing missing data and different statistic methods to combind data
- epiSeeker provides new features for plotCov to plot cluster tree and peak co-accessibility
- epiSeeekr provides interactive function for user to explore data
- epiSeeker provides functions to plot gene structure, base modification and motif
- see more on https://github.com/YuLab-SMU/ChIPseeker/blob/devel/NEWS.md