NxtIRFcore: Core Engine for NxtIRF: a User-Friendly Intron Retention and Alternative Splicing Analysis using the IRFinder Engine

IRFinder

R Documentation

Runs the OpenMP/C++-based NxtIRF/IRFinder algorithm

Description

These function calls the IRFinder C++ routine on one or more BAM files.

The routine is an improved version over the original IRFinder, with OpenMP-based multi-threading and the production of compact "COV" files to record alignment coverage. A NxtIRF reference built using BuildReference is required.

After IRFinder is run, users should call CollateData to collate individual outputs into an experiment / dataset.

BAM2COV creates COV files from BAM files without running the full IRFinder algorithm.

See details for performance info.

Usage

BAM2COV(
  bamfiles = "./Unsorted.bam",
  sample_names = "sample1",
  output_path = "./cov_folder",
  n_threads = 1,
  Use_OpenMP = TRUE,
  overwrite = FALSE,
  verbose = FALSE
)

IRFinder(
  bamfiles = "./Unsorted.bam",
  sample_names = "sample1",
  reference_path = "./Reference",
  output_path = "./IRFinder_Output",
  n_threads = 1,
  Use_OpenMP = TRUE,
  overwrite = FALSE,
  run_featureCounts = FALSE,
  verbose = FALSE
)

Arguments

`bamfiles`	A vector containing file paths of 1 or more BAM files
`sample_names`	The sample names of the given BAM files. Must be a vector of the same length as `bamfiles`
`output_path`	The output directory of this function
`n_threads`	(default `1`) The number of threads to use. See details.
`Use_OpenMP`	(default `TRUE`) Whether to use OpenMP to run IRFinder. If set to `FALSE`, BiocParallel will be used if `n_threads` is set
`overwrite`	(default `FALSE`) If IRFinder output files already exist, will not attempt to re-run. If `run_featureCounts` is `TRUE`, will not overwrite gene counts of previous run unless `overwrite` is `TRUE`.
`verbose`	(default `FALSE`) Set to `TRUE` to allow IRFinder to output progress bars and messages
`reference_path`	The directory containing the NxtIRF reference
`run_featureCounts`	(default `FALSE`) Whether this function will run Rsubread::featureCounts on the BAM files after running IRFinder. If so, the output will be saved to `"main.FC.Rds` in the `output_path` directory as a list object.

Details

Typical run-times for a 100-million paired-end alignment BAM file takes 10 minutes using a single core. Using 8 threads, the runtime is approximately 2 minutes. Approximately 10 Gb of RAM is used when OpenMP is used. If OpenMP is not used (see below), this memory usage is multiplied across the number of processor threads (i.e. 40 Gb if n_threads = 4).

OpenMP is natively available to Linux / Windows compilers, and OpenMP will be used if Use_OpenMP is set to TRUE, using multiple threads to process each BAM file. On Macs, if OpenMP is not available at compilation, BiocParallel will be used, processing BAM files simultaneously, with one BAM file per thread.

Value

IRFinder output will be saved to output_path. Output files will be named using the given sample names.

sample.txt.gz: The main IRFinder output file containing the quantitation of IR and splice junctions, as well as QC information
sample.cov: Contains coverage information in compressed binary. See GetCoverage
main.FC.Rds: A single file containing gene counts for the whole dataset (only if run_featureCounts == TRUE)

Functions

BAM2COV: Converts BAM files to COV files without running IRFinder algorithm
IRFinder: Runs IRFinder algorithm on BAM files. Requires a NxtIRF/IRFinder reference generated by BuildReference()

Examples


# Run BAM2COV, which only produces COV files but does not run IRFinder:

bams <- NxtIRF_example_bams()

BAM2COV(bams$path, bams$sample,
  output_path = file.path(tempdir(), "IRFinder_output"),
  n_threads = 2, overwrite = TRUE
)

# Run IRFinder algorithm, which produces:
# - text output of intron coverage and spliced read counts
# - COV files which record read coverages

example_ref <- file.path(tempdir(), "Reference")

BuildReference(
    reference_path = example_ref,
    fasta = chrZ_genome(),
    gtf = chrZ_gtf()
)

bams <- NxtIRF_example_bams()

IRFinder(bams$path, bams$sample,
  reference_path = file.path(tempdir(), "Reference"),
  output_path = file.path(tempdir(), "IRFinder_output"),
  n_threads = 2
)

alexchwong/NxtIRFcore documentation built on Oct. 31, 2022, 9:14 a.m.