def checkForFiles(self, dirname):
required = ["profile.list", "refpkg", "hmms/concat.hmm", "sets/setinfo.csv"]
for f in required:
p = os.path.join(dirname, f)
if not base.exists(p):
print("Configuartion folder does not contain: {}".format(f))
return False
return True
def readInfo(self, name):
p = os.path.join(self.cfg["db"], "refpkg", name, "CONTENTS.json")
# raise error if we cant find the file
if not base.exists(p):
log("Could not find {}".format(p))
exit(13)
# read and return json
with open(p) as json_file:
j = json.load(json_file)
return j
def pkgfile(self, name, t):
"""
get a file path for a refpkg package
"""
info = self.readInfo(name)
p = os.path.join(self.cfg["db"], "refpkg", name, info["files"][t])
if base.exists(p):
return p
else:
log("Could not find: {}".format(p))
exit(12)
def place(self, fasta, bedfile):
"""
main function to place a bin in the tree.
will subsequently run hmmer
"""
# test if we can open the input files first
if not base.exists(fasta):
logging.error("Could not open fasta file")
self.write_outfile()
exit(1)
if not base.exists(bedfile):
logging.error("Could not open bed file")
self.write_outfile()
exit(1)
# define output files
hmmDir = os.path.join(self.cfg["outdir"], "workfiles", "hmmer")
file.isdir(hmmDir)
hmmOut = os.path.join(hmmDir, "placement.tsv")
hmmOus = os.path.join(hmmDir, "placement.out")
hitOut = os.path.join(hmmDir, "hits.tsv")
# run hmmer if forced or input newer than output
h = hmmer("hmmsearch", fasta, hmmOut, touch=self.cfg["touch"])
if h.doIneedTorun(self.cfg["force"]) or self.cfg["fplace"]:
logging.info("Searching for proteins to place in the tree")
h.run(
hmmOus,
hmmfiles=self.config.placementHMMs,
modus=self.cfg["dbinfo"]["modus"],
evalue=self.cfg["evalue"],
cores=self.cfg["ncores"],
)
# clean hmmer outpout
logging.info("Processing Hmmer results")
hitOut = h.clean(hmmOut, bedfile, hitOut, self.cfg["mindist"])
self.updateStep("findprots", "looked for proteins")
# pplacer paths
placerDir = os.path.join(self.cfg["outdir"], "workfiles", "pplacer")
placerDirTmp = os.path.join(placerDir, "tmp")
pplaceAlinment = os.path.join(placerDir, "horizontalAlignment.fasta")
pplaceOut = os.path.join(placerDir, "placement.jplace")
pplaceLog = os.path.join(placerDir, "placement.log")
pplaceOutReduced = os.path.join(placerDir, "placementReduced.jplace")
file.isdir(placerDirTmp)
# pplacer
logging.debug("Preparing pplacer")
pp = pplacer("pplacer", fasta, pplaceOut, touch=self.cfg["touch"])
if pp.doIneedTorun(self.cfg["force"]) or self.cfg["fplace"]:
logging.debug("Preparing alignments")
pp.prepareAlignment(
pplaceAlinment,
hitOut,
os.path.join(self.cfg["db"], "profile.list"),
fasta,
self.config,
self.cfg,
placerDirTmp,
)
if pp.lenscmgs == 0 and not self.cfg["touch"]:
logging.error("Could not find any marker genes")
self.write_outfile()
exit(1)
else:
logging.info("Placing proteins in tree")
self.updateStep("pplacer", "starting")
pplacer_success = pp.run(
os.path.join(self.cfg["db"], "refpkg", "concat.refpkg"),
logfile=pplaceLog,
cores=self.cfg["ncorespplacer"],
)
if pplacer_success is False:
logging.warning("Pplacer could not finish. Exiting now")
self.write_outfile()
exit(1)
# reduce placements to the placements with at least posterior of p
logging.debug("Reducing placements")
if not self.cfg["touch"]:
pplaceOutReduced = pp.reduceJplace(
pplaceOut, pplaceOutReduced,
self.cfg["minPlacementLikelyhood"])
else:
pp.touch([pplaceOutReduced])
logging.debug("Reducing placements done")
# run TOG to get a tree
togTree = os.path.join(placerDir, "placement.tree")
tg = tog("guppy", pplaceOutReduced, togTree, touch=self.cfg["touch"])
if tg.doIneedTorun(self.cfg["force"]):
logging.debug("Fetching tree")
r = tg.run()
if r is False:
logging.debug("No placement found")
self.write_outfile()
logging.debug("Getting best placements")
# save path to togtree for plotting later
self.cfg["togtreepath"] = togTree
self.cfg["togjson"] = pplaceOutReduced
# now we can place the bin using the tree
if not self.cfg["touch"]:
t = treelineage.treeHandler(togTree, annotate=False)
t2 = treelineage.treeHandler(self.config.tree, annotate=False)
sets = self.getSets()
# get HCA and LCA placements
self.placements = {}
for method in ["LCA", "HPA"]:
self.placements[method] = t.getPlacement(
method,
sets,
t2,
self.cfg["nPlacements"],
self.cfg["minSupport"],
maximum=self.cfg["nEvals"],
debug=self.cfg["debug"],
)
else:
self.placements = {"LCA": "touch", "HCA": "touch"}
logging.info("MAG succesfully placed in tree")
