Iteration error on result of groupby

[daler]

Iteration error on result of groupby

[brentp]

you can probably get around this by adding the full=True kwarg to the groupby method.

you can also iterate over the result of the grouping outside of bedtools as:

for line in open(grouped_bed.fn):
     stuff = line.split("\t")
     ...

[daler]

As Brent points out, full=True does the trick for this case.

Under the hood, there's a bunch of checking to see how many fields there are in a line, if it looks like GFF or not, which fields can be converted to an integer, and so on. Then Interval objects are created with a appropriate chrom/start/stop etc attributes and 0- or 1-based coordinates.

For the general case, I'd prefer not to support arbitrarily-formatted (though created by BEDTools) formats like this. But you could write a converter -- if full=True wasn't an option, then you could modify the last two lines of your example to yield lines that are in fact BED-like:

def reformat(x):
    for line in open(x.fn):
        fields = line.strip().split('\t')
        yield pybedtools.create_interval_from_list([
            fields[1], fields[2], fields[3], fields[0], fields[4], fields[5]])

for line in reformat(grouped_bed):
    print line.fields

[yarden]

Thanks so much Brent and Ryan, I didn't think to use .fn (I guess it's technically only there temporarily right?) and totally missed full=True. Very helpful, thanks!

[daler]

Yep, if you don't do anything special, .fn will be deleted when the interpreter exits. But you could use a .saveas() call if you want to save it for later. Alternatively, when running groupby() (or any method that wraps BEDTools commands), use the output kwarg to have it save the file to a specific filename.