convert hg18 mapping into hg19

[seq_GA]

Hi,

I have set of (huge) data being mapped against to hg18 build. Now I want the mapping to be done against hg19.
Do I have to remap the data against hg19 build? Or do I convert the coordinates using UCSC liftover? Please let me know the details. Thx.

[ffinkernagel]

It depends. The easy way is of course using UCSC liftover (you'll need the .chain file and liftOver to run it locally on a linux system), but you might learn something from remapping - perhaps some data that had not been mappable previously now is.

Here's a python function that takes a list like [(chr, start, stop)], runs liftover using a given chain file and get's you another list with the new coordinates.

    def do_liftover(listOfChromosomeIntervals, chain_file):
        """perform a lift over. Error messages are silently swallowed!"""
        tmp_input = tempfile.NamedTemporaryFile()
        tmp_output = tempfile.NamedTemporaryFile()
        tmp_error = tempfile.NamedTemporaryFile()
        max_len = 0
        for row in listOfChromosomeIntervals:
            tmp_input.write(" ".join(str(x) for x in row))
            tmp_input.write("\n")
            max_len = max(len(row), max_len)
        tmp_input.write("\n")
        tmp_input.flush()#it's magic ;)
        cmd = [os.path.join(os.path.dirname(__file__), 'chains','liftOver'), tmp_input.name, chain_file, 
               tmp_output.name, tmp_error.name]
        p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr = subprocess.PIPE)
        p.communicate()
        tmp_output.seek(0, os.SEEK_SET)
        res = []
        for row in tmp_output:
            row = row.strip().split("\t")
            row[1] = int(row[1])
            row[2] = int(row[2])
            res.append(tuple(row))
        tmp_error.seek(0, os.SEEK_SET)
        d = tmp_error.read()
        tmp_input.close()
        tmp_output.close()
        tmp_error.close()
        return res

[seq_GA]

Ok ffinkernagel. Thx for your response.

[lmf_bill]

I think the best way is to remap against hg19. It will take the similar time when you use liftover to transfer hg18 to hg19. Good lucky!

[simonandrews]

I think the best way is to remap against hg19. It will take the similar time when you use liftover to transfer hg18 to hg19. Good lucky!

That's not been our experience. It may be worth remapping against the newer assembly to get the most from your data, but running liftOver is a much quicker option. LiftOver is ridiculously fast!

[lmf_bill]

yes, If you only want to get the location transfer, liftover is faster. In the mean time, many novel cases need to be consider, just as you said you can get more info from the newer assembly. Besides, some additional assemble regions will also make it difficult. So, the easy way is remapping. By the way, how huge are your data? several terabyte?