def readMetadata(metaFile):
metadata = meta.Metadata(metaFile)
plates = metadata.data["Metadata_Plate"].unique()
print("Total plates:",len(plates))
for i in range(len(plates)):
#plate = metadata.filterRecords(lambda df: (df.Metadata_Plate == plates[0]) & (df.Metadata_Well == "a01"), copy=True)
plate = metadata.filterRecords(lambda df: (df.Metadata_Plate == plates[i]), copy=True)
yield plate
return
def readPlates(metaFile):
metadata = meta.Metadata(metaFile)
plates = metadata.data["Metadata_Plate"].unique()
utils.logger.info("Total plates: " + str(len(plates)))
for i in range(len(plates)):
plate = metadata.filterRecords(lambda df:
(df.Metadata_Plate == plates[i]),
copy=True)
yield plate
return
def readDataset(metaFile, images_dir):
# Read metadata and split data in training and validation
metadata = meta.Metadata(metaFile, dtype=None)
trainingFilter = lambda df: df["Allele_Replicate"] <= 5
validationFilter = lambda df: df["Allele_Replicate"] > 5
metadata.splitMetadata(trainingFilter, validationFilter)
# Create a dataset
keyGen = lambda r: "{}/{}-{}".format(r["Metadata_Plate"], r[
"Metadata_Well"], r["Metadata_Site"])
dataset = ds.Dataset(metadata, "Allele", CHANNELS, images_dir, keyGen)
print(metadata.data.iloc[100])
return dataset
def processMetadata(plate_maps, barcode_file, csv_list, root):
# Load plate maps data and create labels
plateMaps = meta.Metadata(plate_maps, "multi", "blanks")
maps = plateMaps.data
maps["Treatment"] = maps["broad_sample"] + "@" + maps[
"mmoles_per_liter"].astype(str)
maps["Compound"] = 0
treatments = maps["Treatment"].unique()
compounds = maps["broad_sample"].unique()
print("Unique treatments:", len(treatments))
for i in range(len(treatments)):
maps.loc[lambda df: df.Treatment == treatments[i], "Treatment"] = i
utils.printProgress(i + 1, len(treatments), prefix="Treatments")
print("Unique compounds:", len(compounds))
for i in range(len(compounds)):
maps.loc[lambda df: df.broad_sample == compounds[i], "Compound"] = i
utils.printProgress(i + 1, len(compounds), prefix="Compounds")
# Load barcodes and csv files
barcodes = meta.Metadata(barcode_file, "single")
load_data = meta.Metadata(csv_list, "multi")
# Merge two frames: csvs + barcodes to attach compound layout to each image
columns = list(load_data.data.columns.values)
metadata = pd.merge(load_data.data.drop(columns[13:], axis=1),
barcodes.data,
left_on=["Metadata_Plate"],
right_on=["Assay_Plate_Barcode"],
how="inner")
metadata = metadata.drop(
["Batch_Number", "Batch_Date", "Assay_Plate_Barcode"], axis=1)
del load_data, barcodes
# Concatenate paths and filenames and make them relative
metadata = relativePaths(metadata, "RNA", "PathName_OrigRNA",
"FileName_OrigRNA", root)
metadata = relativePaths(metadata, "ER", "PathName_OrigER",
"FileName_OrigER", root)
metadata = relativePaths(metadata, "AGP", "PathName_OrigAGP",
"FileName_OrigAGP", root)
metadata = relativePaths(metadata, "Mito", "PathName_OrigMito",
"FileName_OrigMito", root)
metadata = relativePaths(metadata, "DNA", "PathName_OrigDNA",
"FileName_OrigDNA", root)
print(metadata.info())
# Merge two frames: metadata + plateMaps to attach treatment info to each image
metadata = pd.merge(metadata,
maps,
left_on=["Plate_Map_Name", "Metadata_Well"],
right_on=["plate_map_name", "well_position"],
how="left")
metadata = metadata.drop([
"plate_map_name", "well_position", "broad_sample", "mg_per_ml",
"mmoles_per_liter", "solvent"
],
axis=1)
metadata[
"plate_well"] = metadata["Metadata_Plate"] + metadata["Metadata_Well"]
# Find replicate labels
metadata["Treatment_Replicate"] = 0
replicateDistribution = {}
for i in range(len(treatments)):
mask1 = metadata["Treatment"] == i
wells = metadata[mask1]["plate_well"].unique()
utils.printProgress(i + 1, len(treatments), "Replicates")
replicate = 1
for j in range(len(wells)):
mask2 = metadata["plate_well"] == wells[j]
metadata.loc[mask1 & mask2, "Treatment_Replicate"] = replicate
replicate += 1
try:
replicateDistribution[replicate - 1] += 1
except:
replicateDistribution[replicate - 1] = 1
metadata = metadata.drop(["plate_well"], axis=1)
print(replicateDistribution)
print(metadata.info())
# Save resulting metadata
metadata.to_csv("metadata.csv", index=False)
dframe = pd.DataFrame({
"ID": pd.Series(range(len(treatments))),
"Treatment": pd.Series(treatments)
})
dframe.to_csv("treatments.csv", index=False)
dframe = pd.DataFrame({
"ID": pd.Series(range(len(compounds))),
"Compound": pd.Series(compounds)
})
dframe.to_csv("compounds.csv", index=False)
def processMetadata(plate_maps, barcode_file, csv_list, root):
# Load plate maps data and create labels
plateMaps = meta.Metadata(plate_maps, "multi", "tabs")
maps = plateMaps.data
maps["Allele"] = maps["NCBIGeneID"].astype(
str) + "@" + maps["x_mutation_status"]
maps["Gene"] = 0
alleles = maps["Allele"].unique()
genes = maps["NCBIGeneID"].unique()
print("Unique alleles:", len(alleles))
for i in range(len(alleles)):
maps.loc[lambda df: df.Allele == alleles[i], "Allele"] = i
utils.printProgress(i + 1, len(alleles), prefix="Alleles")
print("Unique genes:", len(genes))
for i in range(len(genes)):
maps.loc[lambda df: df.broad_sample == genes[i], "Gene"] = i
utils.printProgress(i + 1, len(genes), prefix="Genes")
# Load barcodes and csv files
barcodes = meta.Metadata(barcode_file, "single")
load_data = meta.Metadata(csv_list, "multi")
# Merge two frames: csvs + barcodes to attach gene layout to each image
columns = list(load_data.data.columns.values)
metadata = pd.merge(load_data.data.drop(columns[13:], axis=1),
barcodes.data,
left_on=["Metadata_Plate"],
right_on=["Assay_Plate_Barcode"],
how="inner")
del load_data, barcodes
# Concatenate paths and filenames and make them relative
metadata = relativePaths(metadata, "RNA", "PathName_OrigRNA",
"FileName_OrigRNA", root)
metadata = relativePaths(metadata, "ER", "PathName_OrigER",
"FileName_OrigER", root)
metadata = relativePaths(metadata, "AGP", "PathName_OrigAGP",
"FileName_OrigAGP", root)
metadata = relativePaths(metadata, "Mito", "PathName_OrigMito",
"FileName_OrigMito", root)
metadata = relativePaths(metadata, "DNA", "PathName_OrigDNA",
"FileName_OrigDNA", root)
print(metadata.info())
# Merge two frames: metadata + plateMaps to attach treatment info to each image
metadata = pd.merge(metadata,
maps,
left_on=["Plate_Map_Name", "Metadata_Well"],
right_on=["plate_map_name", "well_position"],
how="left")
metadata = metadata.drop([
"plate_map_name", "well_position", "broad_sample", "NCBIGeneID",
"pert_type", "PublicID", "Transcript", "VirusPlateName",
"well_position", "x_mutation_status", "broad_sample", "pert_name"
],
axis=1)
metadata["plate_well"] = metadata["Metadata_Plate"].astype(
str) + "::" + metadata["Metadata_Well"]
# Find replicate labels
metadata["Allele_Replicate"] = 0
replicateDistribution = {}
for i in range(len(alleles)):
mask1 = metadata["Allele"] == i
wells = metadata[mask1]["plate_well"].unique()
utils.printProgress(i + 1, len(alleles), "Replicates")
replicate = 1
for j in range(len(wells)):
mask2 = metadata["plate_well"] == wells[j]
metadata.loc[mask1 & mask2, "Allele_Replicate"] = replicate
replicate += 1
try:
replicateDistribution[replicate - 1] += 1
except:
replicateDistribution[replicate - 1] = 1
metadata = metadata.drop(["plate_well"], axis=1)
print(replicateDistribution)
print(metadata.info())
# Save resulting metadata
metadata.to_csv("metadata.csv", index=False)
dframe = pd.DataFrame({
"ID": pd.Series(range(len(alleles))),
"Allele": pd.Series(alleles)
})
dframe.to_csv("alleles.csv", index=False)
dframe = pd.DataFrame({
"ID": pd.Series(range(len(genes))),
"Gene": pd.Series(genes)
})
dframe.to_csv("genes.csv", index=False)
