tximeta
leverages the hashed digest of the Salmon or piscem index,
in addition to a number of core Bioconductor packages (GenomicFeatures,
ensembldb, AnnotationHub, GenomeInfoDb, BiocFileCache) to automatically
populate metadata for the user, without additional effort from the user.
For other quantifiers see the customMetaInfo
argument below.
tximeta(
coldata,
type = NULL,
txOut = TRUE,
skipMeta = FALSE,
skipSeqinfo = FALSE,
useHub = TRUE,
markDuplicateTxps = FALSE,
cleanDuplicateTxps = FALSE,
customMetaInfo = NULL,
skipFtp = FALSE,
...
)
a data.frame with at least two columns (others will propogate to object):
files
- character, paths of quantification files
names
- character, sample names
if coldata
is a vector, it is assumed to be the paths of quantification files
and unique sample names are created
what quantifier was used (see tximport
)
whether to output transcript-level data.
tximeta
is designed to have transcript-level output
with Salmon, so default is TRUE
,
and it's recommended to use summarizeToGene
following tximeta
for gene-level summarization.
For an alevin file, tximeta
will import the
gene level counts ignoring this argument (alevin
produces only gene-level quantification).
whether to skip metadata generation
(e.g. to avoid errors if not connected to internet).
This calls tximport
directly and so either
txOut=TRUE
or tx2gene
should be specified.
whether to skip the addition of Seqinfo, which requires an internet connection to download the relevant chromosome information table from UCSC
whether to first attempt to download a TxDb/EnsDb
object from AnnotationHub, rather than creating from a
GTF file from FTP (default is TRUE). If FALSE, it will
force tximeta
to download and parse the GTF
whether to mark the status
(hasDuplicate
) and names of duplicate transcripts
(duplicates
) in the rowData of the SummarizedExperiment output.
Subsequent summarization to gene level will keep track
of the number of transcripts sets per gene (numDupSets
)
whether to try to clean duplicate transcripts (exact sequence duplicates) by replacing the transcript names that do not appear in the GTF with those that do appear in the GTF
the relative path to a custom metadata
information JSON file, relative to the paths in files
of
coldata
. For example, customMetaInfo="meta_info.json"
would indicate that in the same directory as the quantification
files in files
, there are custom metadata information
JSON files. These should contain the SHA-256 hash of the
reference transcripts with the index_seq_hash
tag
(see details in vignette).
whether to avoid ftp://
in case of
firewall, default is FALSE
arguments passed to tximport
a SummarizedExperiment with metadata on the rowRanges
.
(if the hashed digest in the Salmon or Sailfish index does not match
any known transcriptomes, or any locally saved linkedTxome
,
tximeta
will just return a non-ranged SummarizedExperiment)
Most of the code in tximeta
works to add metadata and transcript ranges
when the quantification was performed with Salmon. However,
tximeta
can be used with any quantification type
that is supported
by tximport
, where it will return an non-ranged SummarizedExperiment.
tximeta
performs a lookup of the hashed digest of the index
(stored in an auxilary information directory of the Salmon output)
against a database of known transcriptomes, which lives within the tximeta
package and is continually updated on Bioconductor's release schedule.
In addition, tximeta
performs a lookup of the digest against a
locally stored table of linkedTxome
's (see link{makeLinkedTxome}
).
If tximeta
detects a match, it will automatically populate,
e.g. the transcript locations, the transcriptome release,
the genome with correct chromosome lengths, etc. It allows for automatic
and correct summarization of transcript-level quantifications to the gene-level
via summarizeToGene
without the need to manually build
a tx2gene
table.
tximeta
on the first run will ask where the BiocFileCache for
this package should be kept, either using a default location or a temporary
directory. At any point, the user can specify a location using
setTximetaBFC
and this choice will be saved for future sessions.
Multiple users can point to the same BiocFileCache, such that
transcript databases (TxDb or EnsDb) associated with certain Salmon indices
and linkedTxomes
can be accessed by different users without additional
effort or time spent downloading and building the relevant TxDb / EnsDb.
Note that, if the TxDb or EnsDb is present in AnnotationHub, tximeta
will
use this object instead of downloading and building a TxDb/EnsDb from GTF
(to disable this set useHub=FALSE).
In order to allow that multiple users can read and write to the same location, one should set the BiocFileCache directory to have group write permissions (g+w).
# point to a Salmon quantification file:
dir <- system.file("extdata/salmon_dm", package="tximportData")
files <- file.path(dir, "SRR1197474", "quant.sf")
coldata <- data.frame(files, names="SRR1197474", condition="A", stringsAsFactors=FALSE)
# normally we would just run the following which would download the appropriate metadata
# se <- tximeta(coldata)
# for this example, we instead point to a local path where the GTF can be found
# by making a linkedTxome:
indexDir <- file.path(dir, "Dm.BDGP6.22.98_salmon-0.14.1")
fastaFTP <- c("ftp://ftp.ensembl.org/pub/release-98/fasta/drosophila_melanogaster/cdna/Drosophila_melanogaster.BDGP6.22.cdna.all.fa.gz",
"ftp://ftp.ensembl.org/pub/release-98/fasta/drosophila_melanogaster/ncrna/Drosophila_melanogaster.BDGP6.22.ncrna.fa.gz")
gtfPath <- file.path(dir, "Drosophila_melanogaster.BDGP6.22.98.gtf.gz")
makeLinkedTxome(indexDir=indexDir, source="LocalEnsembl", organism="Drosophila melanogaster",
release="98", genome="BDGP6.22", fasta=fastaFTP, gtf=gtfPath, write=FALSE)
#> linkedTxome is same as already in bfc
se <- tximeta(coldata)
#> importing quantifications
#> reading in files with read.delim (install 'readr' package for speed up)
#> 1
#>
#> found matching linked transcriptome:
#> [ LocalEnsembl - Drosophila melanogaster - release 98 ]
#> loading existing TxDb created: 2024-10-18 15:14:14
#> loading existing transcript ranges created: 2024-10-18 15:14:14
#> Warning:
#>
#> Warning: the annotation is missing some transcripts that were quantified.
#> 5 out of 33706 txps were missing from GTF/GFF but were in the indexed FASTA.
#> (This occurs sometimes with Ensembl txps on haplotype chromosomes.)
#> In order to build a ranged SummarizedExperiment, these txps were removed.
#> To keep these txps, and to skip adding ranges, use skipMeta=TRUE
#>
#> Example missing txps: [FBtr0307759, FBtr0084079, FBtr0084080, ...]
# to clear the entire linkedTxome table
# (don't run unless you want to clear this table!)
# bfcloc <- getTximetaBFC()
# bfc <- BiocFileCache(bfcloc)
# bfcremove(bfc, bfcquery(bfc, "linkedTxomeTbl")$rid)