Contig to chromosome mapping with Ensembl

ensembl-test.py

import requests
import sys

def ensembl_get(ext):
    uri = 'http://rest.ensemblgenomes.org' + ext
    r = requests.get(uri,
                 headers = { "Content-Type": "application/json" })

    if not r.ok:
        r.raise_for_status()
    return r.json()

def ensembl_assembly(species):
    return ensembl_get('/info/assembly/{}?'.format(species))

def ensembl_genome_assembly(assembly_id):
    return ensembl_get('/info/genomes/assembly/{}?'.format(assembly_id))

def ensembl_assembly_region(species, region):
    return ensembl_get('/info/assembly/{}/{}?'.format(species, region))

def ensembl_map(species, source_asm, source_region,
                target_asm, coord_system,
                target_coord_system):
    ext = '/map/{}/{}/{}/{}?'.format(species, source_asm,
                                          source_region, target_asm)
    ext += 'coord_system={}&target_coord_system={}'.format(coord_system, target_coord_system)
    
    return ensembl_get(ext)

def ensembl_map_contig_to_chromosome(species, asm, region):
    return ensembl_map(species, asm, region, asm, 'contig', 'chromosome')

##################

def remap(contig):
    info = ensembl_map_contig_to_chromosome('hordeum_vulgare', 'ASM32608v1', contig)
    if len(info['mappings']) != 1:
        raise 'More than one mapping for {}!'.format(contig)
    mapping = info['mappings'][0]['mapped']
    
    chromosome = mapping['seq_region_name']
    start = mapping['start']
    end = mapping['end']
    print('{}\t{}:{}-{}'.format(contig,chromosome,start,end))
    
for contig in sys.argv[1:]:
    remap(contig)