Specifying genomes

This section illustrates the use of genome files for use with BEDTools programs that need to know chromosome limits to prevent out-of-range coordinates.

Using BEDTools programs like slopBed or shuffleBed from the command line requires “genome” or “chromsizes” files. pybedtools comes with common genome assemblies already set up as a dictionary with chromosomes as keys and zero-based (start, stop) tuples as values:

>>> from pybedtools import genome_registry
>>> genome_registry.dm3['chr2L']
(0, 23011544)

The rules for specifying a genome for methods that require a genome are as follows (use whatever is most convenient):

  • Use g to specify either a filename or a dictionary

  • Use genome to specify either an assembly name or a dictionary

Below are examples of each.

As a file

This is the typical way of using BEDTools programs, by specifying an existing genome file with g:

>>> a = pybedtools.example_bedtool('a.bed')
>>> b = a.slop(b=100, g='hg19.genome')

As a string

This is probably the most convenient way of specifying a genome. If the genome exists in the genome registry it will be used directly. Alternatively, if you have `genomepy<https://github.com/vanheeringen-lab/genomepy/>`_ installed, you can use the genomepy genome name, such as hg38. In this case, the genome file will be located automatically. Finally, if the genome is not in the registry or managed by genomepy, it will automatically be downloaded from UCSC. You must use the genome kwarg for this; if you use g a string will be interpreted as a filename:

>>> c = a.slop(b=100, genome='hg19')

As a dictionary

This is a good way of providing custom coordinates; either g or genome will accept a dictionary:

>>> d = a.slop(b=100, g={'chr1':(1, 10000)})
>>> e = a.slop(b=100, genome={'chr1':(1,100000)})

Make sure that all these different methods return the same results

>>> b == c == d == e
True

Converting to a file

Since BEDTools programs operate on files, the fastest choice will be to use an existing file. While the time to convert a dictionary to a file is extremely small, over 1000’s of files (e.g., for Monte Carlo simulations), the time may add up. The function pybedtools.chromsizes_to_file() will create a file from a dictionary or string:

>>> # with no filename specified, a tempfile will be created
>>> pybedtools.chromsizes_to_file(pybedtools.chromsizes('dm3'), 'dm3.genome')
'dm3.genome'
>>> print(open('dm3.genome').read())
chr2L       23011544
chr2LHet    368872
chr2R       21146708
chr2RHet    3288761
chr3L       24543557
chr3LHet    2555491
chr3R       27905053
chr3RHet    2517507
chr4        1351857
chrM        19517
chrU        10049037
chrUextra   29004656
chrX        22422827
chrXHet     204112
chrYHet     347038