srFilter: Functions for user-created and built-in ShortRead filters

In Bioconductor/ShortRead: FASTQ input and manipulation

Functions for user-created and built-in ShortRead filters

Description

These functions create user-defined (srFitler) or built-in instances of SRFilter objects. Filters can be applied to objects from ShortRead, returning a logical vector to be used to subset the objects to include only those components satisfying the filter.

Usage

srFilter(fun, name = NA_character_, ...)
## S4 method for signature 'missing'
srFilter(fun, name=NA_character_, ...)
## S4 method for signature 'function'
srFilter(fun, name=NA_character_, ...)

compose(filt, ..., .name)

idFilter(regex=character(0), fixed=FALSE, exclude=FALSE,
         .name="idFilter")
occurrenceFilter(min=1L, max=1L,
                 withSread=c(NA, TRUE, FALSE),
                 duplicates=c("head", "tail", "sample", "none"),
                 .name=.occurrenceName(min, max, withSread,
                                       duplicates))
nFilter(threshold=0L, .name="CleanNFilter")
polynFilter(threshold=0L, nuc=c("A", "C", "T", "G", "other"),
           .name="PolyNFilter")
dustyFilter(threshold=Inf, batchSize=NA, .name="DustyFilter")
srdistanceFilter(subject=character(0), threshold=0L,
                 .name="SRDistanceFilter")

##
## legacy filters for ungapped alignments
##

chromosomeFilter(regex=character(0), fixed=FALSE, exclude=FALSE,
                 .name="ChromosomeFilter")
positionFilter(min=-Inf, max=Inf, .name="PositionFilter")
strandFilter(strandLevels=character(0), .name="StrandFilter")
alignQualityFilter(threshold=0L, .name="AlignQualityFilter")
alignDataFilter(expr=expression(), .name="AlignDataFilter")

Arguments

fun	An object of class function to be used as a filter. fun must accept a single named argument x, and is expected to return a logical vector such that x[fun(x)] selects only those elements of x satisfying the conditions of fun
name	A character(1) object to be used as the name of the filter. The name is useful for debugging and reference.
filt	A SRFilter object, to be used with additional arguments to create a composite filter.
.name	An optional character(1) object used to over-ride the name applied to default filters.
regex	Either character(0) or a character(1) regular expression used as grep(regex, chromosome(x)) to filter based on chromosome. The default (character(0)) performs no filtering
fixed	logical(1) passed to grep, influencing how pattern matching occurs.
exclude	logical(1) which, when TRUE, uses regex to exclude, rather than include, reads.
min	numeric(1)
max	numeric(1). For positionFilter, min and max define the closed interval in which position must be found min <= position <= max. For occurrenceFilter, min and max define the minimum and maximum number of times a read occurs after the filter.
strandLevels	Either character(0) or character(1) containing strand levels to be selected. ShortRead objects have standard strand levels NA, "+", "-", "*", with NA meaning strand information not available and "*" meaning strand information not relevant.
withSread	A logical(1) indicating whether uniqueness includes the read sequence (withSread=TRUE), is based only on chromosome, position, and strand (withSread=FALSE), or only the read sequence (withSread=NA), as described for occurrenceFilter below..
duplicates	Either character{1}, a function name, or a function taking a single argument. Influence how duplicates are handled, as described for occurrenceFilter below.
threshold	A numeric(1) value representing a minimum (srdistanceFilter, alignQualityFilter) or maximum (nFilter, polynFilter, dustyFilter) criterion for the filter. The minima and maxima are closed-interval (i.e., x >= threshold, x <= threshold for some property x of the object being filtered).
nuc	A character vector containing IUPAC symbols for nucleotides or the value "other" corresponding to all non-nucleotide symbols, e.g., N.
batchSize	NA or an integer(1) vector indicating the number of DNA sequences to be processed simultaneously by dustyFilter. By default, all reads are processed simultaneously. Smaller values use less memory but are computationally less efficient.
subject	A character() of any length, to be used as the corresponding argument to srdistance.
expr	A expression to be evaluated with pData(alignData(x)).
...	Additional arguments for subsequent methods; these arguments are not currently used.

Details

srFilter allows users to construct their own filters. The fun argument to srFilter must be a function accepting a single argument x and returning a logical vector that can be used to select elements of x satisfying the filter with x[fun(x)]

The signature(fun="missing") method creates a default filter that returns a vector of TRUE values with length equal to length(x).

compose constructs a new filter from one or more existing filter. The result is a filter that returns a logical vector with indices corresponding to components of x that pass all filters. If not provided, the name of the filter consists of the names of all component filters, each separated by " o ".

The remaining functions documented on this page are built-in filters that accept an argument x and return a logical vector of length(x) indicating which components of x satisfy the filter.

idFilter selects elements satisfying grep(regex, id(x), fixed=fixed).

chromosomeFilter selects elements satisfying grep(regex, chromosome(x), fixed=fixed).

positionFilter selects elements satisfying min <= position(x) <= max.

strandFilter selects elements satisfying match(strand(x), strand, nomatch=0) > 0.

occurrenceFilter selects elements that occur >=min and <=max times. withSread determines how reads will be treated: TRUE to include the sread, chromosome, strand, and position when determining occurrence, FALSE to include chromosome, strand, and position, and NA to include only sread. The default is withSread=NA. duplicates determines how reads with more than max reads are treated. head selects the first max reads of each set of duplicates, tail the last max reads, and sample a random sample of max reads. none removes all reads represented more than max times. The user can also provide a function (as used by tapply) of a single argument to select amongst reads.

nFilter selects elements with fewer than threshold 'N' symbols in each element of sread(x).

polynFilter selects elements with fewer than threshold copies of any nucleotide indicated by nuc.

dustyFilter selects elements with high sequence complexity, as characterized by their dustyScore. This emulates the dust command from WindowMaker software. Calculations can be memory intensive; use batchSize to process the argument to dustyFilter in batches of the specified size.

srdistanceFilter selects elements at an edit distance greater than threshold from all sequences in subject.

alignQualityFilter selects elements with alignQuality(x) greater than threshold.

alignDataFilter selects elements with pData(alignData(x)) satisfying expr. expr should be formulated as though it were to be evaluated as eval(expr, pData(alignData(x))).

Value

srFilter returns an object of SRFilter.

Built-in filters return a logical vector of length(x), with TRUE indicating components that pass the filter.

Author(s)

Martin Morgan <[e-mail]>

See Also

SRFilter.

Examples

sp <- SolexaPath(system.file("extdata", package="ShortRead"))
aln <- readAligned(sp, "s_2_export.txt") # Solexa export file, as example

# a 'chromosome 5' filter
filt <- chromosomeFilter("chr5.fa")
aln[filt(aln)]
# filter during input
readAligned(sp, "s_2_export.txt", filter=filt)

# x- and y- coordinates stored in alignData, when source is SolexaExport
xy <- alignDataFilter(expression(abs(x-500) > 200 & abs(y-500) > 200))
aln[xy(aln)]

# both filters as a single filter
chr5xy <- compose(filt, xy)
aln[chr5xy(aln)]

# both filters as a collection
filters <- c(filt, xy)
subsetByFilter(aln, filters)
summary(filters, aln)

# read, chromosome, strand, position tuples occurring exactly once
aln[occurrenceFilter(withSread=TRUE, duplicates="none")(aln)]
# reads occurring exactly once
aln[occurrenceFilter(withSread=NA, duplicates="none")(aln)]
# chromosome, strand, position tuples occurring exactly once
aln[occurrenceFilter(withSread=FALSE, duplicates="none")(aln)]

# custom filter: minimum calibrated base call quality >20
goodq <- srFilter(function(x) {
    apply(as(quality(x), "matrix"), 1, min, na.rm=TRUE) > 20
}, name="GoodQualityBases")
goodq
aln[goodq(aln)]

Bioconductor/ShortRead documentation built on Nov. 2, 2024, 4:38 p.m.