def dGenomeMutAdd(project, mutID, mutfasta):
'''
Check and add a mutant
'''
if not os.path.exists(mutfasta):
logger.error('Fasta file %s may not be present'%(mutfasta))
return False
org = Organism(project)
if not org.isOrg(mutID):
logger.warning('Organism %s is not present yet!'%mutID)
return False
gen = Genome(project)
gen.addProteome(mutID, mutfasta)
logger.info('Mutant %s (%s) added, having %d mutated genes'
%(mutID, org.getOrg(mutID).mkind,gen.howMany(mutID)))
return True
def dGenomeAdd(project, orgID, filename):
'''
Add a single genome
'''
if not os.path.exists(filename):
logger.error('Fasta file %s may not be present'%(filename))
return False
filename = os.path.abspath(filename)
org = Organism(project)
if not org.isOrg(orgID):
logger.warning('Organism %s is not present yet!'%orgID)
return False
gen = Genome(project)
gen.addProteome(orgID, filename)
logger.info('Added genome %s, having %d proteins'%
(orgID, gen.howMany(orgID)))
return True
def dGenomeStats(project, svg=False, doPrint=True):
# Which project are we talking about?
kind = dSetKind(project)
proj = Project(project)
organism = Organism(project)
genome = Genome(project)
kegg = Kegg(project)
if kind == 'single' or kind == 'pangenome':
logger.info('Single genomes stats')
# Single genomes stats
# Header
header = '\t'.join( ['ID', 'name', 'description', 'proteome size',
'mapped to kegg', 'KEGG orthology IDs',
'pathways', 'reactions'] )
if doPrint:
print header
else:
logger.info(header)
lOrg = []
for org in organism.getAll():
org_id = org.org_id
name = org.name if org.name else 'NONE'
description = org.description if org.description else 'NONE'
prots = genome.howMany(org_id)
mapped, ko, react, path = (kegg.howManyMapped(org_id),
kegg.howManyKO(org_id),
kegg.howManyReactions(org_id),
kegg.howManyPathways(org_id))
stats = '\t'.join( [str(x) for x in [org_id, name, description,
prots, mapped, ko, path,
react]] )
if doPrint:
print stats
else:
logger.info(stats)
lOrg.append([org_id, prots, mapped, react])
plotMapBars(lOrg, 'Single genomes statistics', 'single', svg)
if proj.isPanGenome():
logger.info('Pangenome stats')
# Pangenome stats
# Header
header = '\t'.join( ['kind', 'size',
'mapped to kegg', 'KEGG orthology IDs',
'pathways', 'reactions'] )
if doPrint:
print header
else:
logger.info(header)
core, acc, uni = (genome.getLenCore(), genome.getLenAcc(),
genome.getLenUni())
stats = []
stats.append('\t'.join( [str(x) for x in ['core', core,
kegg.howManyMapped(pangenome='core'),
kegg.howManyKO(pangenome='core'),
kegg.howManyPathways(pangenome='core'),
kegg.howManyReactions(pangenome='core')]]))
stats.append('\t'.join( [str(x) for x in ['accessory', acc,
kegg.howManyMapped(pangenome='accessory'),
kegg.howManyKO(pangenome='accessory'),
kegg.howManyPathways(pangenome='accessory'),
kegg.howManyReactions(pangenome='accessory')]]))
stats.append('\t'.join( [str(x) for x in ['unique', uni,
kegg.howManyMapped(pangenome='unique'),
kegg.howManyKO(pangenome='unique'),
kegg.howManyPathways(pangenome='unique'),
kegg.howManyReactions(pangenome='unique')]]))
for stat in stats:
if doPrint:
print stat
else:
logger.info(stat)
lPanGenome = [['Core', core, kegg.howManyMapped(pangenome='core'),
kegg.howManyReactions(pangenome='core')],
['Accessory', acc,
kegg.howManyMapped(pangenome='accessory'),
kegg.howManyReactions(pangenome='accessory')],
['Unique', uni,
kegg.howManyMapped(pangenome='unique'),
kegg.howManyReactions(pangenome='unique')]]
plotMapBars(lPanGenome, 'PanGenome statistics', 'pangenome_stats',
svg)
plotPanGenome(core, acc, uni, svg)
elif kind == 'mutants':
refs = [org.org_id
for org in organism.getAll()
if not organism.isMutant(org.org_id)]
# Header
header = '\t'.join( ['ID', 'name', 'description', 'kind', 'proteome size',
'mapped to kegg', 'reactions'] )
for ref_id in refs:
logger.info('Mutants of %s stats'%ref_id)
if doPrint:
print header
else:
logger.info(header)
muts = [x for x in organism.getOrgMutants(ref_id)]
lOrg = []
for org_id in [ref_id] + muts:
org = organism.getOrg(org_id)
name = org.name if org.name else 'NONE'
description = org.description if org.description else 'NONE'
mkind = org.mkind if org.mkind in ['deletion', 'insertion'] else 'wild-type'
if mkind not in ['deletion', 'insertion']:
prots = genome.howMany(org_id)
elif mkind == 'deletion':
prots = genome.howMany(ref_id) - genome.howMany(org_id)
elif mkind == 'insertion':
prots = genome.howMany(ref_id) + genome.howMany(org_id)
mapped, react = (kegg.howManyMapped(org_id),
kegg.howManyReactions(org_id))
if mkind == 'deletion':
mapped = kegg.howManyMapped(ref_id) - mapped
react = kegg.howManyReactions(ref_id) - react
elif mkind == 'insertion':
mapped += kegg.howManyMapped(ref_id)
react += kegg.howManyReactions(ref_id)
stats = '\t'.join( [str(x) for x in [org_id, name, description,
mkind, prots, mapped,
react]] )
if doPrint:
print stats
else:
logger.info(stats)
lOrg.append([org_id, prots, mapped, react])
plotMapBars(lOrg, 'Wild-type (%s) and mutants statistics'%ref_id,
'%s'%ref_id, svg)
else:
logger.info('No statistics can be computed at this time')
return False
return True