make_genomiccoord: Generate sliding-window genomic coordinate strings for Tm...

In TmCalculator: Genome-Wide Nucleic Acid Melting Temperature Profiling and Multi-Omics Integration

Generate sliding-window genomic coordinate strings for Tm calculation

Description

Tiles one or more chromosomes from a BSgenome object into overlapping windows and returns a named character vector of coordinate strings in the format "chr:start-end:strand:genome_pkg:region_id". This vector is the primary input for downstream Tm calculation functions (tm_nn, tm_gc, tm_wallace) that accept genomic coordinate strings.

Usage

make_genomiccoord(
  bsgenome,
  chromosomes = NULL,
  window = 200L,
  slide = 50L,
  start = NULL,
  end = NULL,
  strand = "+",
  trim_N = c("ends", "filter", "none"),
  max_N_frac = 0.1,
  N_scan_block = window,
  region_prefix = "region",
  genome_pkg_name = NULL,
  as_vector = TRUE,
  verbose = TRUE
)

Arguments

bsgenome	A BSgenome object (e.g. BSgenome.Hsapiens.UCSC.hg38) or a character string with the BSgenome package name (loaded automatically).
chromosomes	Character vector of chromosome names to tile. Must be present in seqlevels(bsgenome). Default: the 24 standard human chromosomes paste0("chr", c(1:22, "X", "Y")).
window	Integer. Width of each sliding window in base pairs. Default 200L.
slide	Integer. Step size between consecutive window starts in base pairs. slide == window gives non-overlapping tiling; slide < window gives overlapping windows. Default 50L.
start	Integer or NULL. Override the start position for every chromosome. When NULL (default) the start is chromosome position 1 (adjusted for N-trimming if trim_N != "none"). Can be a named integer vector to set different starts per chromosome.
end	Integer or NULL. Override the end position for every chromosome. When NULL (default) the end is the chromosome length (adjusted for N-trimming). Can be a named integer vector.
strand	Character. Strand label embedded in the coordinate string. One of "+" (default) or "-".
trim_N	Character. N-base handling strategy. One of "ends" (default), "filter", or "none". See section N-base trimming above.
max_N_frac	Numeric in [0, 1]. Maximum fraction of N bases tolerated per window. Windows exceeding this threshold are dropped. Only used when trim_N = "filter". Default 0.1.
N_scan_block	Integer. Block size (bp) for the coarse N-end detection scan used by trim_N = "ends". Larger values are faster but less precise. Default 10000L.
region_prefix	Character. Prefix for region IDs embedded in the coordinate string. Default "region" (producing region1, region2, ...).
genome_pkg_name	Character or NULL. The genome package name to embed in the coordinate string (4th field). When NULL (default), the name is extracted automatically from the bsgenome object via S4Vectors::metadata(bsgenome). Supply explicitly when using a custom BSgenome object whose metadata name differs from the canonical package name.
as_vector	Logical. When TRUE (default), return a plain named character vector. When FALSE, return a data.frame with columns coord, chr, start, end, strand, genome, region_id, chr_start_used, chr_end_used – useful for downstream GRanges construction.
verbose	Logical. Print per-chromosome progress messages. Default TRUE.

Value

When as_vector = TRUE (default): a named character vector of coordinate strings, one element per window. Names are the region IDs (region1, region2, ...).

When as_vector = FALSE: a data.frame with columns coord, chr, win_start, win_end, strand, genome, region_id, chr_start_used, chr_end_used.

Coordinate string format

Each element of the returned vector follows the pattern:

  chr1:10001-10200:+:BSgenome.Hsapiens.UCSC.hg38:region1
  chr1:10051-10250:+:BSgenome.Hsapiens.UCSC.hg38:region2
  ...

Fields (colon-separated):

1. chromosome – e.g. chr1

2. start-end – 1-based, inclusive coordinates

3. strand – + or -

4. genome – BSgenome package name (character)

5. region_id – unique label regionN

N-base trimming

Human (and most mammalian) chromosomes begin and end with long stretches of N bases that represent assembly gaps. Windows that consist entirely or predominantly of Ns produce meaningless Tm values. The function offers two trimming strategies controlled by trim_N:

"none"

No trimming. Windows start at start and end at end (or chromosome length).

"ends"

Detect the first and last positions of non-N bases on each chromosome using Biostrings::letterFrequency() on a coarse block scan, then trim start and end to those positions. Efficient: reads the chromosome sequence only once in blocks.

"filter"

Generate windows across the full start-end range but then remove any window whose N-base fraction exceeds max_N_frac (default 0.1, i.e. 10%). More granular than "ends" but slower because it reads each window sequence.

Author(s)

Junhui Li

See Also

tm_nn, tm_gc, tm_wallace, BSgenome, letterFrequency

Examples

## Not run: 
library(BSgenome.Hsapiens.UCSC.hg38)

## -- Basic usage: tile chr1 with 200 bp windows, 50 bp slide ----------
coords <- make_genomiccoord(
  bsgenome    = BSgenome.Hsapiens.UCSC.hg38,
  chromosomes = "chr1",
  window      = 200L,
  slide       = 50L
)
length(coords)      # number of windows on chr1
head(coords, 3)
# region1 "chr1:10001-10200:+:BSgenome.Hsapiens.UCSC.hg38:region1"
# region2 "chr1:10051-10250:+:BSgenome.Hsapiens.UCSC.hg38:region2"
# region3 "chr1:10101-10300:+:BSgenome.Hsapiens.UCSC.hg38:region3"

## -- Non-overlapping tiling (slide == window) -------------------------
coords_nonoverlap <- make_genomiccoord(
  bsgenome    = BSgenome.Hsapiens.UCSC.hg38,
  chromosomes = paste0("chr", 1:22),
  window      = 200L,
  slide       = 200L    # no overlap
)
length(coords_nonoverlap)   # ~15 million windows across autosomes

## -- Custom start/end (e.g. a specific sub-region) --------------------
coords_sub <- make_genomiccoord(
  bsgenome    = BSgenome.Hsapiens.UCSC.hg38,
  chromosomes = "chr1",
  window      = 200L,
  slide       = 50L,
  start       = 1000000L,
  end         = 2000000L
)
length(coords_sub)   # 19,981 windows in 1 Mb region

## -- No N-trimming (use full chromosome length) ------------------------
coords_noN <- make_genomiccoord(
  bsgenome    = BSgenome.Hsapiens.UCSC.hg38,
  chromosomes = "chr1",
  window      = 200L,
  slide       = 50L,
  trim_N      = "none"
)

## -- Per-window N-filtering (removes windows with >10% N) -------------
coords_filt <- make_genomiccoord(
  bsgenome    = BSgenome.Hsapiens.UCSC.hg38,
  chromosomes = "chr1",
  window      = 200L,
  slide       = 50L,
  trim_N      = "filter",
  max_N_frac  = 0.10
)

## -- Get data.frame output for GRanges construction -------------------
df <- make_genomiccoord(
  bsgenome    = BSgenome.Hsapiens.UCSC.hg38,
  chromosomes = "chr1",
  window      = 200L,
  slide       = 50L,
  as_vector   = FALSE
)
gr <- GenomicRanges::GRanges(
  seqnames = df$chr,
  ranges   = IRanges::IRanges(start = df$win_start, end = df$win_end),
  strand   = df$strand
)

## -- Pass directly to tm_nn -------------------------------------------
coords <- make_genomiccoord(
  bsgenome    = BSgenome.Hsapiens.UCSC.hg38,
  chromosomes = "chr1",
  window      = 200L,
  slide       = 200L,
  start       = 1000000L,
  end         = 1010000L
)
tm_results <- tm_nn(coords, Na = 50)

## End(Not run)

TmCalculator documentation built on Aug. 28, 2026, 5:09 p.m.