R Interface

R interface

Run these examples from the FusionPep project root after completing setup. The project-local R environment must be active. FusionPep is currently a script-based R project; load its functions with source().

Analyze the bundled inputs

for (path in list.files("R", pattern = "[.]R$", full.names = TRUE)) source(path)
source("report/fusion_report.R")

result <- run_fusion_analysis(
  sequence_file = "input/sequences.fasta",
  peptide_file = "input/peptides.csv",
  junction_file = "input/fusion_junctions.csv"
)

result$peptide_summary
result$junction_evaluations
result$coverage

run_fusion_analysis() reads and validates inputs, normalizes peptides, maps sequences, evaluates supplied junctions, calculates coverage, and aligns the fusion with each parent. It returns a fusion_peptide_mapping_result list without writing output files.

Analysis arguments

The public analysis function accepts:

Argument	Meaning	Default
sequence_file	Protein FASTA path	Required
peptide_file	Peptide CSV path	Required
sequence_ids	Named mapping from roles to FASTA identifiers	Fusion, ParentA, ParentB
peptide_column	Peptide sequence column	"peptide"
il_equivalent	Match I and L equivalently	TRUE
junction_file	Explicit junction CSV path	NULL
gap_opening	Pairwise alignment gap-opening penalty	10
gap_extension	Pairwise alignment gap-extension penalty	0.5
substitution_matrix	Requested alignment matrix	"BLOSUM62"
store_alignment_objects	Retain raw alignment S4 objects	FALSE

For custom FASTA identifiers, keep the role names Fusion, ParentA, and ParentB in sequence_ids and set their values to the respective identifiers in your file. Junction CSV role fields still use those canonical role names.

With junction_file = NULL, mapping and alignment proceed and fusion matches have an unassessed junction state. Unlike the CLI, this function does not default to the bundled junction file.

I/L equivalence controls peptide matching. The alignment output records exact and I/L-equivalent identity separately. Alignment summaries and the run manifest record the matrix actually used; inspect those values when interpreting alignments containing non-standard amino-acid symbols.

Returned data

The result retains config, selected reference records, the original peptide table, normalized input rows, validated junctions, mapped occurrences, junction evaluations, peptide summaries, coverage, uncovered intervals, alignment tables, warnings, and the run manifest.

Raw alignment objects are omitted by default to reduce retained memory. Complete alignment columns, regions, and summaries remain available. Set store_alignment_objects = TRUE only when your downstream R code needs the raw S4 objects.

Write outputs separately

Using the result created above:

output_dir <- file.path(tempdir(), "fusionpep-example")
paths <- write_fusion_outputs(result, output_dir)
paths$report <- write_fusion_report(result, output_dir, paths)
paths$report

This example uses the R session's temporary directory. Choose a persistent output directory to retain the files after the session ends. Reusing a directory replaces generated files with matching names.

write_fusion_outputs() writes CSV, RDS, warning, and figure files and returns their paths. The returned result keeps full input paths; the saved RDS and manifest record only the input file names and hashes. write_fusion_report() consumes the result and those paths, writes HTML, and returns the report path. Figure PDFs require Cairo graphics. Reports and figures describes the saved files.

Reuse mapping or alignment without figure export

The lower-level functions let a script request only the calculations it needs:

• read_fusion_sequences() returns the selected sequences, headers, requested identifiers, source path, and input hash.
• read_peptide_table() returns the original input, normalized rows, selected column, source path, and input hash.
• read_fusion_junctions() returns validated junction rows and their source information; pass the selected sequences for coordinate checks.
• map_peptides_to_references() takes selected sequences, normalized rows, optional validated junction rows, and the I/L setting. It returns occurrence, evaluation, summary, coverage, and uncovered-interval tables.
• align_fusion_to_parents() returns the pairwise alignment columns, regions, summaries, and optional raw objects.

Use the readers to prepare inputs for these lower-level calls. Keep the same il_equivalent setting when normalizing and mapping. The mapping and alignment functions do not write files or create figures.

R function paths resolve relative to the current R working directory. The CLI's project-root path resolution does not apply to arbitrary R calls.