When iterating through bed_file: "Error: The requested bed file (<file>) could not be opened. Exiting!"

[daler]

When iterating through bed_file: "Error: The requested bed file () could not be opened. Exiting!"

from pybedtools.

[daler]

Hmm, I'll need to look into this more. But you can accomplish it in a different way, which should be much faster too:

Usually when modifying features you'd write a modifying function and give it to BedTool.each(). But in this case, since you want to rename using which line you're on, I think it's easiest to use a generator:

def generator():
    for i, feature in enumerate(bed):
        x.name = "GSM_{0}".format(i+1)
        yield x

pybedtools.BedTool(generator()).saveas("renamed.bed")

Keep in mind though if you don't already have a "name" field, you'll need to use pybedtools.featurefuncs.extend_fields(x, 4) before setting the name in the generator.

[alanlhutchison]

Hi,

Thank you for your speedy response.

I tried the generator trick you showed me. I think I do not understand it fully. I am attempting to run these on a test file TestD.bed, which I generate thusly
echo 'chr2L\t20\t30\tTestD1\nchr2L\t45\t60\tTestD2\n' > TestD.bed

Here is what I attempted as far as code goes:

import pybedtools
from pybedtools import BedTool

def gen_change_name(bed,GSM):
for i, feature in enumerate(bed):
feature.name = GSM+"_{0}".format(i+1)
yield feature
#BedTool(gen_change_name(bed,GSM)).saveas("temp.bed")

def change_name(bed,fn):
GSM = st.split(fn,'_')[0]
mod4= gen_change_name(bed,GSM)
i=0
for line in mod4:
print line
print bed[i]
bed[i] = line #this produces an error "TypeError: 'BedTool' object does not support item assignment"
i+=1
return bed

I can now generate a generator that will spit out the "corrected" lines, but I cannot pass them to the bed files that are already there. It seems if I call BedTool on my generator, it returns a generator and not a bed file, and I do not get the desired output. How can I get the generator output into a BedTool?

Also, GSM is just a string, nothing more.

If my query has turned into one about generators and requires its own thread I apologize.

Please advise...

[daler]

Hmm, it's difficult to see exactly what you're doing here . . . can you edit your response so that the code is wrapped in triple backticks (``` -- see http://github.github.com/github-flavored-markdown/) with the correct whitespace?

[alanlhutchison]

Hi,

Sorry about that! Thank you for the very helpful link! I hope my question is more understandable now. Thank you as well for your patience and speedy response.

I tried the generator code you showed me. I think I do not understand it fully. I am attempting to run these on a test file TestD.bed, which I generate thusly

echo 'chr2L\t20\t30\tTestD1\nchr2L\t45\t60\tTestD2\n' > TestD.bed

and produces:

cat TestD.bed
chr2L   20  30  TestD1
chr2L   45  60  TestD2

Here is what I attempted as far as code goes:

import pybedtools
from pybedtools import BedTool

def gen_change_name(bed,GSM):
    #GSM is just a string
    for i, feature in enumerate(bed):
        feature.name = GSM+"_{0}".format(i+1)
        yield feature
    #BedTool(gen_change_name(bed,GSM)).saveas("temp.bed") 
    #The above line produces an error

def change_name(bed,fn):
    GSM = st.split(fn,'_')[0]
    mod4= gen_change_name(bed,GSM)
    i=0
    for line in mod4:
        print line
        print bed[i]
        bed[i] = line #this produces an error "TypeError: 'BedTool' object does not support item assignment"
        i+=1
    return bed

I can now generate a generator that will spit out the "corrected" lines, but I cannot pass them to the bed files that are already there. It seems if I call BedTool on my generator, it returns a generator and not a bed file, and I do not get the desired output. How can I get the generator output into a BedTool?

Also, GSM is just a string, nothing more.

If my query has turned into one about generators and requires its own thread I apologize.

Please advise...

[daler]

Here is a self-contained script that creates the test files. Based on how you're getting GSM, I'm assuming there's some meaningful information in the filenames that you want to get, so I changed the example filenames to reflect that.

You can play around with this test file to help figure out how it works and adapt it to your needs.

import pybedtools

# Create two example files so that this example is self-contained.
pybedtools.BedTool('''
chr2L 20 30 TestD1
chr2L 45 60 TestD2''', from_string=True).saveas('a_test.bed')

pybedtools.BedTool('''
chr3L 500 600 TestB1
chr3L 900 1000 TestB2''', from_string=True).saveas('b_test.bed')


def gen_change_name(fn, GSM):
    '''
    This generator accepts a filename and a string, and yields
    pybedtools.Interval features with names changed according to `GSM` and line
    number.
    '''
    for i, feature in enumerate(pybedtools.BedTool(fn)):
        feature.name = GSM + '_{0}'.format(i + 1)
        yield feature


def change_name(fn):
    '''
    This function accepts a filename and creates a new file with changed names.
    It returns a BedTool pointing to this new file.
    '''
    GSM = fn.split('_')[0]

    # This is the key: BedTool objects can be created from a generator of
    # pybedtools.Interval objects...which is what gen_name_change was designed
    # to do.
    return pybedtools.BedTool(gen_change_name(fn, GSM))\
        .saveas(fn + '.changed')

for fn in ('a_test.bed', 'b_test.bed'):
    original = pybedtools.BedTool(fn)
    changed = change_name(fn)
    print 'original', fn
    print original
    print 'changed', fn
    print changed

[alanlhutchison]

Hi daler,

Thanks a lot! This works great! I appreciate your time!