megadepth quick start guide
7 May 2023
Package
megadepth 1.10.0
The goal of megadepth is to provide an R interface to the command line tool Megadepth for BigWig and BAM related utilities created by Christopher Wilks (Wilks, Ahmed, Baker, Zhang, Collado-Torres, and Langmead, 2020). This R package enables fast processing of BigWig files on downstream packages such as dasper and recount3. The Megadepth software also provides utilities for processing BAM files and extracting coverage information from them.
Here is an illustration on how fast megadepth is compared to other tools for processing local and remote BigWig files.
Figure 1: Timing results
Timing comparison for processing 1,000 genomic regions on a bigWig file that is available on the local disk or on a remote resource. We compared megadepth against rtracklayer and pyBigWig. megadepth is typically faster that these other software solutions for computing the mean coverage across a set of 1,000 input genomic regions. Check https://github.com/LieberInstitute/megadepth/tree/master/analysis for more details.
Throughout the documentation we use a capital M to refer to the software by Christopher Wilks and a lower case m to refer to this R/Bioconductor package.
1 Basics
1.1 Install megadepth
R is an open-source statistical environment which can be easily modified to enhance its functionality via packages. megadepth is a R package available via the Bioconductor repository for packages. R can be installed on any operating system from CRAN after which you can install megadepth by using the following commands in your R session:
if (!requireNamespace("BiocManager", quietly = TRUE)) {
install.packages("BiocManager")
}
BiocManager::install("megadepth")
## Check that you have a valid Bioconductor installation
BiocManager::valid()1.2 Required knowledge
megadepth is based on many other packages and in particular in those that have implemented the infrastructure needed for dealing with RNA-seq data and high-throughput sequencing data in general. You might benefit from being familiar with the BigWig file format and the rtracklayer for importing those files into R as well as exporting BED files (Lawrence, Gentleman, and Carey, 2009). If you are working with annoation files, GenomicFeatures and GenomicRanges will also be useful to you.
If you are asking yourself the question “Where do I start using Bioconductor?” you might be interested in this blog post.
1.3 Asking for help
As package developers, we try to explain clearly how to use our packages and in which order to use the functions. But R and Bioconductor have a steep learning curve so it is critical to learn where to ask for help. The blog post quoted above mentions some but we would like to highlight the Bioconductor support site as the main resource for getting help: remember to use the megadepth tag and check the older posts. Other alternatives are available such as creating GitHub issues and tweeting. However, please note that if you want to receive help you should adhere to the posting guidelines. It is particularly critical that you provide a small reproducible example and your session information so package developers can track down the source of the error.
1.4 Citing megadepth
We hope that megadepth will be useful for your research. Please use the following information to cite the package and the overall approach. Thank you!
## Citation info
citation("megadepth")
#> To cite package 'megadepth' in publications use:
#>
#> Zhang D, Collado-Torres L (2023). _megadepth: BigWig and BAM related
#> utilities_. doi:10.18129/B9.bioc.megadepth
#> <https://doi.org/10.18129/B9.bioc.megadepth>,
#> https://github.com/LieberInstitute/megadepth - R package version
#> 1.10.0, <http://www.bioconductor.org/packages/megadepth>.
#>
#> Wilks C, Ahmed O, Baker DN, Zhang D, Collado-Torres L, Langmead B
#> (2020). "Megadepth: efficient coverage quantification for BigWigs and
#> BAMs." _bioRxiv_. doi:10.1101/2020.12.17.423317
#> <https://doi.org/10.1101/2020.12.17.423317>,
#> <https://www.biorxiv.org/content/10.1101/2020.12.17.423317v1>.
#>
#> To see these entries in BibTeX format, use 'print(<citation>,
#> bibtex=TRUE)', 'toBibtex(.)', or set
#> 'options(citation.bibtex.max=999)'.2 Quick start
To get started, we need to load the megadepth package into our R session. This will load all the required dependencies.
library("megadepth")Once we have the R package loaded, we need to install the Megadepth software. We can do so with install_megadepth(), which downloads a binary for your OS (Linux, Windows or macOS) 1 Please check Megadepth for instructions on how to compile the software from source if the binary version doesn’t work for you.. We can then use with an example BigWig file to compute the coverage at a set of regions.
## Install the latest version of Megadepth
install_megadepth(force = TRUE)
#> The latest megadepth version is 1.2.0
#> This is not an interactive session, therefore megadepth has been installed temporarily to
#> /tmp/Rtmpr8Akxv/megadepthNext, we might want to use megadepth to quantify the coverage at a set of regions of the genome of interest to us. Here we will use two example files that are include in the package for illustration and testing purposes. One of them is a bigWig file that contains the base-pair coverage information for a sample of interest and the second one is BED file which contains the genomic region coordinates of interest. So we first locate them.
## Next, we locate the example BigWig and annotation files
example_bw <- system.file("tests", "test.bam.all.bw",
package = "megadepth", mustWork = TRUE
)
annotation_file <- system.file("tests", "testbw2.bed",
package = "megadepth", mustWork = TRUE
)
## Where are they?
example_bw
#> [1] "/tmp/RtmpepljC4/Rinst2c8aa62f8927b7/megadepth/tests/test.bam.all.bw"
annotation_file
#> [1] "/tmp/RtmpepljC4/Rinst2c8aa62f8927b7/megadepth/tests/testbw2.bed"Once we have located the example files we can proceed to calculating the base-pair coverage for our genomic regions of interest. There are several ways to do this with megadepth, but here we use the user-friendly function get_coverage(). This function will perform a given operation op on the bigWig file for a given set of regions of interest (annotation). One of those operations is to compute the mean base-pair coverage for each input region. This is what we’ll do with our example bigWig file.
## We can then use megadepth to compute the coverage
bw_cov <- get_coverage(
example_bw,
op = "mean",
annotation = annotation_file
)
bw_cov
#> GRanges object with 4 ranges and 1 metadata column:
#> seqnames ranges strand | score
#> <Rle> <IRanges> <Rle> | <numeric>
#> [1] chr10 0-10 * | 0.00
#> [2] chr10 8756697-8756762 * | 15.85
#> [3] chr10 4359156-4359188 * | 3.00
#> [4] GL000219.1 168500-168620 * | 1.26
#> -------
#> seqinfo: 2 sequences from an unspecified genome; no seqlengthsget_coverage() returns an object that is familiar to GenomicRanges users, that is, a GRanges object that can be used with other Bioconductor software packages (Huber, Carey, Gentleman, Anders, Carlson, Carvalho, Bravo, Davis, Gatto, Girke, Gottardo, Hahne, Hansen, Irizarry, Lawrence, Love, MacDonald, Obenchain, Oleś, Pagès, Reyes, Shannon, Smyth, Tenenbaum, Waldron, and Morgan, 2015).
This example is just the tip of the iceberg, as Megadepth and thus megadepth can do a lot of useful processing operations on BAM and bigWig files.
3 Users guide
3.1 Command interface
Megadepth is very powerful and can do a lot of different things. The R/Bioconductor package provides two functions for interfacing with Megadepth, megadepth_cmd() and megadepth_shell(). For the first one, megadepth_cmd(), you need to know the actual command syntax you want to use and format it accordingly. If you are more comfortable with R functions, megadepth_shell() uses cmdfun to power this interface and capture the standard output stream into R.
To make it easier to use, megadepth includes functions that simplify the number of arguments, read in the output files, and converts them into R/Bioconductor friendly objects, such as get_coverage() illustrated previously in the quick start section.
We hope that you’ll find megadepth and Megadepth useful for your work. If you are interested in checking how fast megadepth is, check out the speed analysis comparison against other tools. Note that the size of the files used and the number of genomic regions queried will affect the speed comparisons.
## R-like interface
## that captures the standard output into R
head(megadepth_shell(help = TRUE))
#> [1] "megadepth 1.2.0" ""
#> [3] "BAM and BigWig utility." ""
#> [5] "Usage:" " megadepth <bam|bw|-> [options]"
## Command-like interface
megadepth_cmd("--help")#> megadepth 1.2.0
#>
#> BAM and BigWig utility.
#>
#> Usage:
#> megadepth <bam|bw|-> [options]
#>
#> Options:
#> -h --help Show this screen.
#> --version Show version.
#> --threads # of threads to do: BAM decompression OR compute sums over multiple BigWigs in parallel
#> if the 2nd is intended then a TXT file listing the paths to the BigWigs to process in parallel
#> should be passed in as the main input file instead of a single BigWig file (EXPERIMENTAL).
#> --prefix String to use to prefix all output files.
#> --no-auc-stdout Force all AUC(s) to be written to <prefix>.auc.tsv rather than STDOUT
#> --no-annotation-stdout Force summarized annotation regions to be written to <prefix>.annotation.tsv rather than STDOUT
#> --no-coverage-stdout Force covered regions to be written to <prefix>.coverage.tsv rather than STDOUT
#> --keep-order Output annotation coverage in the order chromosomes appear in the BAM/BigWig file
#> The default is to output annotation coverage in the order chromosomes appear in the annotation BED file.
#> This is only applicable if --annotation is used for either BAM or BigWig input.
#>
#> BigWig Input:
#> Extract regions and their counts from a BigWig outputting BED format if a BigWig file is detected as input (exclusive of the other BAM modes):
#> Extracts all reads from the passed in BigWig and output as BED format.
#> This will also report the AUC over the annotated regions to STDOUT.
#> If only the name of the BigWig file is passed in with no other args, it will *only* report total AUC to STDOUT.
#> --annotation <bed> Only output the regions in this BED applying the argument to --op to them.
#> --op <sum[default], mean, min, max> Statistic to run on the intervals provided by --annotation
#> --sums-only Discard coordinates from output of summarized regions
#> --distance (2200[default]) Number of base pairs between end of last annotation and start of new to consider in the same BigWig query window (a form of binning) for performance. This determines the number of times the BigWig index is queried.
#> --unsorted (off[default]) There's a performance improvement *if* BED file passed to --annotation is 1) sorted by sort -k1,1 -k2,2n (default is to assume sorted and check for unsorted positions, if unsorted positions are found, will fall back to slower version)
#> --bwbuffer <1GB[default]> Size of buffer for reading BigWig files, critical to use a large value (~1GB) for remote BigWigs.
#> Default setting should be fine for most uses, but raise if very slow on a remote BigWig.
#>
#>
#> BAM Input:
#> Extract basic junction information from the BAM, including co-occurrence
#> If only the name of the BAM file is passed in with no other args, it will *only* report total AUC to STDOUT.
#> --fasta Path to the reference FASTA file if a CRAM file is passed as the input file (ignored otherwise)
#> If not passed, references will be downloaded using the CRAM header.
#> --junctions Extract co-occurring jx coordinates, strand, and anchor length, per read
#> writes to a TSV file <prefix>.jxs.tsv
#> --all-junctions Extract all jx coordinates, strand, and anchor length, per read for any jx
#> writes to a TSV file <prefix>.all_jxs.tsv
#> --longreads Modifies certain buffer sizes to accommodate longer reads such as PB/Oxford.
#> --filter-in Integer bitmask, any bits of which alignments need to have to be kept (similar to samtools view -f).
#> --filter-out Integer bitmask, any bits of which alignments need to have to be skipped (similar to samtools view -F).
#> --add-chr-prefix Adds "chr" prefix to relevant chromosomes for BAMs w/o it, pass "human" or "mouse".
#> Only works for human/mouse references (default: off).
#>
#> Non-reference summaries:
#> --alts Print differing from ref per-base coverages
#> Writes to a CSV file <prefix>.alts.tsv
#> --include-softclip Print a record to the alts CSV for soft-clipped bases
#> Writes total counts to a separate TSV file <prefix>.softclip.tsv
#> --only-polya If --include-softclip, only print softclips which are mostly A's or T's
#> --include-n Print mismatch records when mismatched read base is N
#> --print-qual Print quality values for mismatched bases
#> --delta Print POS field as +/- delta from previous
#> --require-mdz Quit with error unless MD:Z field exists everywhere it's
#> expected
#> --head Print sequence names and lengths in SAM/BAM header
#>
#> Coverage and quantification:
#> --coverage Print per-base coverage (slow but totally worth it)
#> --auc Print per-base area-under-coverage, will generate it for the genome
#> and for the annotation if --annotation is also passed in
#> Defaults to STDOUT, unless other params are passed in as well, then
#> if writes to a TSV file <prefix>.auc.tsv
#> --bigwig Output coverage as BigWig file(s). Writes to <prefix>.bw
#> (also <prefix>.unique.bw when --min-unique-qual is specified).
#> Requires libBigWig.
#> --annotation <BED|window_size> Path to BED file containing list of regions to sum coverage over
#> (tab-delimited: chrm,start,end). Or this can specify a contiguous region size in bp.
#> --op <sum[default], mean> Statistic to run on the intervals provided by --annotation
#> --no-index If using --annotation, skip the use of the BAM index (BAI) for pulling out regions.
#> Setting this can be faster if doing windows across the whole genome.
#> This will be turned on automatically if a window size is passed to --annotation.
#> --min-unique-qual <int>
#> Output second bigWig consisting built only from alignments
#> with at least this mapping quality. --bigwig must be specified.
#> Also produces second set of annotation sums based on this coverage
#> if --annotation is enabled
#> --double-count Allow overlapping ends of PE read to count twice toward
#> coverage
#> --num-bases Report total sum of bases in alignments processed (that pass filters)
#> --gzip Turns on gzipping of coverage output (no effect if --bigwig is passsed),
#> this will also enable --no-coverage-stdout.
#>
#> Other outputs:
#> --read-ends Print counts of read starts/ends, if --min-unique-qual is set
#> then only the alignments that pass that filter will be counted here
#> Writes to 2 TSV files: <prefix>.starts.tsv, <prefix>