def cacheDBscanClusters(cfg): cacheFileName = dbscan.getDBscanClustersCacheFileName(cfg) if os.path.isfile(cacheFileName): return isolates = pyroprinting.loadIsolates(cfg) clusters = dbscan.computeDBscanClusters(isolates, cfg) with open(cacheFileName, mode='w+b') as cacheFile: pickle.dump(clusters, cacheFile)
def cacheNeighbors(cfg):
cacheFileName = fullsearch.getNeighborsMapCacheFileName(cfg)
if os.path.isfile(cacheFileName):
return
isolates = pyroprinting.loadIsolates(cfg)
neighbors = fullsearch.computeNeighborsMap(isolates, cfg)
avgMatches = sum(len(matches) for _, matches in neighbors.items()) / len(neighbors)
print("avgMatches: {}".format(avgMatches))
with open(cacheFileName, mode='w+b') as cacheFile:
pickle.dump(neighbors, cacheFile)
def loadFromCSV(filename, outfile):
cfg = config.loadConfig()
isolates = pyroprinting.loadIsolates(cfg)
isolateIdMap = {iso.name.strip(): iso for iso in isolates}
clusters = []
with open(filename) as csvFile:
csvLines = "".join(line for line in csvFile if line.strip()).splitlines(True) # remove blank lines
# print(csvLines)
csvReader = csv.reader(csvLines, delimiter=',')
pastHeader = False # because Aldrin's csv files have some header rows
currentClusterId = None
currentCluster = None
for i, row in enumerate(csvReader):
# print("{}/{}".format(i+1, len(csvLines)))
if row[0] == "Cluster Id":
pastHeader = True
elif pastHeader:
if row[0].startswith("Threshold:") or row[0] == "******":
print("Multiple clusterings detected in file. Skipping the rest.")
break
isoId = row[1].strip()
if isoId in isolateIdMap:
if row[0] != currentClusterId:
currentClusterId = row[0]
currentCluster = set()
clusters.append(currentCluster)
currentCluster.add(isolateIdMap[isoId])
else:
print("extra isolate: {}".format(isoId))
# print(clusters)
print(len(clusters))
with open(outfile, mode='w+b') as cacheFile:
pickle.dump(clusters, cacheFile)
len([size for size in clusterSizes if size >= 128])
)
noise = [isolate for isolate in isolates if isolate not in clusterMap]
return count, minmax, (mean, math.sqrt(var)), len(noise), sizeHistogram
def filterSingletonClusters(clusters):
filtered = [cluster for cluster in clusters if len(cluster) > 1]
print("{}/{}".format(len(filtered), len(clusters)))
return filtered
if __name__ == '__main__':
cfg = config.loadConfig()
assert cfg.isolateSubsetSize == "Shared"
isolates = pyroprinting.loadIsolates(cfg)
# dbscanClusters = dbscan.getDBscanClusters(isolates, cfg)
dbscan1Clusters = dbscan.loadDBscanClustersFromFile("dbscanShared_0.995_0.995_1.pickle")
dbscan3Clusters = dbscan.loadDBscanClustersFromFile("dbscanShared_0.995_0.995_3.pickle")
ohclust99Clusters = filterSingletonClusters(importClusters.getOHClustClusters(99))
# ohclust995Clusters = filterSingletonClusters(importClusters.getOHClustClusters(995))
agglomerativeClusters = filterSingletonClusters(importClusters.getAgglomerativeClusters())
replicatePearsons = loadReplicatePearsons(cfg)
db1Pair = ("DBSCAN 1", dbscan1Clusters)
db3Pair = ("DBSCAN 3", dbscan3Clusters)
oh99Pair = ("OHClust 99", ohclust99Clusters)
# oh995Pair = ("OHClust 995", ohclust995Clusters)
aggPair = ("AGGLOMERATIVE", agglomerativeClusters)