TmCalculator - Genome-Wide Nucleic Acid Melting Temperature Profiling and
Multi-Omics Integration
Accurate calculation of nucleic acid melting temperature
(Tm) is fundamental to many molecular biology applications, and
this software scales Tm analysis from individual sequences to
genome‑wide thermodynamic profiling. This package extends Tm
analysis from simple sequence level computation to
comprehensive genome-wide thermodynamic profiling. It takes
four input sources: sequence strings, a FASTA file, an
installed 'BSgenome' package named by string, or a 'GRanges'
carrying sequences. A 'regions' argument selects what to cover
and 'window' and 'slide' set the resolution at which it is
tiled. The implementation provides three Tm calculation
methods: the Wallace rule (Thein & Wallace, 1986), empirical
GC‑content formulas (Marmur, 1962; Schildkraut, 1965; Wetmur,
1991; Untergasser, 2012; von Ahsen, 2001), and nearest‑neighbor
thermodynamics (Breslauer, 1986; Sugimoto, 1996; Allawi, 1997,
1998; SantaLucia, 2004; Freier, 1986; Xia, 1998; Chen, 2012;
Bommarito, 2000; Turner, 2010; Sugimoto, 1995; Peyret, 1999;
SantaLucia & Peyret, 2001; Watkins & SantaLucia, 2005; Zuber,
2022; Ghosh, 2020, 2023). Nearest-neighbor parameter sets are
provided for DNA, RNA and RNA/DNA hybrid duplexes. These
include sets obtained by melting-temperature optimization that
are fitted directly at a stated sodium concentration (Weber,
2015; Ferreira, 2019; Basilio Barbosa, 2019; Banerjee, 2020),
which replace salt correction rather than being corrected; salt
correction is skipped automatically when the requested
condition matches the one a set was fitted at. The Zuber (2022)
set additionally replaces the single terminal-AU penalty with
end terms that depend on the penultimate base pair, applied
automatically at both duplex ends. Parameter sets measured
under molecular crowding (Ghosh, 2020, 2023) are also provided
for DNA and RNA duplexes, so that duplex stability can be
evaluated under cell-like rather than dilute-solution
conditions. Salt corrections are otherwise applied to the
nearest-neighbor model (SantaLucia, 1996, 1998; Owczarzy, 2004,
2008), while each empirical GC-content formula carries the salt
term it was published with; corrections for chemical conditions
such as dimethyl sulfoxide and formamide apply to both. A
compiled C++ core, and task partitioning by region across
'BiocParallel' workers through a 'BPPARAM' argument, profile
the human genome in 3 minutes on a six-core laptop. This
package returns result as a GRanges object for interoperability
with Bioconductor workflows and downstream multi-omics
analyses. Data-level integration reconciles Tm windows with
external multi-omics GRanges objects through overlap,
nearest-feature, windowed-count, and binned-average strategies,
returning a single unified GRanges object ready for downstream
analysis. Visualization-level integration renders multiple
feature layers as independent concentric tracks on a shared
genomic axis, each retaining its native coordinate resolution.
Group comparison supports Wilcoxon rank-sum and Student's
t-tests with multiple available correction methods for
contrasting Tm and other features across region classes.
