edit

GFF3

Description

GFF3 is a text-based file format for representing genomic annotations. The major difference from BED is that GFF3 is more structured and can include sequences in the FASTA file format.

I/O tools for GFF3 are provided from the GenomicFeatures.GFF3 module, which exports following three types:

Reader type: GFF3.Reader
Writer type: GFF3.Writer
Element type: GFF3.Record

A GFF3 file may contain directives and/or comments in addition to genomic features. These lines are skipped by default but you can control the behavior by passing keyword arguments to GFF3.Reader. See the docstring for details.

Examples

Here is a common workflow to iterate over all records in a GFF3 file:

# Import the GFF3 module.
using GenomicFeatures

# Open a GFF3 file.
reader = open(GFF3.Reader, "data.gff3")

# Iterate over records.
for record in reader
    # Do something on record (see Accessors section).
    seqid = GFF3.seqid(reader)
    # ...
end

# Finally, close the reader
close(reader)

If you are interested in directives (which starts with '#') in addition to genomic features, you need to pass skip_directives=false when initializing a GFF3 constructor:

# Set skip_directives to true (this is set to false by default).
reader = GFF3.Reader(open("data.gff3"), skip_directives=false)
for record in record
    # Branch by record type.
    if GFF3.isfeature(record)
        # ...
    elseif GFF3.isdirective(record)
        # ...
    else
        # This never happens.
        assert(false)
    end
end
close(reader)

GenomicFeatures.jl supports tabix to retrieve records overlapping with a specific interval. First you need to create a block compression file from a GFF3 file using bgzip and then index it using tabix.

cat data.gff3 | grep -v "^#" | sort -k1,1 -k4,4n | bgzip >data.gff3.bgz
tabix data.gff3.bgz  # this creates data.gff3.bgz.tbi

Then you can read the block compression file as follows:

# Read the block compression gzip file.
reader = GFF3.Reader("data.gff3.bgz")
for record in eachoverlap(reader, Interval("chr1", 250_000, 300_000))
    # Each record overlap the query interval.
    # ...
end

API

# GenomicFeatures.GFF3.Reader — Type.

GFF3.Reader(input::IO;
            index=nothing,
            save_directives::Bool=false,
            skip_features::Bool=false,
            skip_directives::Bool=true,
            skip_comments::Bool=true)

GFF3.Reader(input::AbstractString;
            index=:auto,
            save_directives::Bool=false,
            skip_features::Bool=false,
            skip_directives::Bool=true,
            skip_comments::Bool=true)

Create a reader for data in GFF3 format.

The first argument specifies the data source. When it is a filepath that ends with .bgz, it is considered to be block compression file format (BGZF) and the function will try to find a tabix index file (.tbi) and read it if any. See http://www.htslib.org/doc/tabix.html for bgzip and tabix tools.

Arguments

input: data source (IO object or filepath)
index: path to a tabix file
save_directives: flag to save directive records (which can be accessed with GFF3.directives)
skip_features: flag to skip feature records
skip_directives: flag to skip directive records
skip_comments: flag to skip comment records