def __init__(self, gene_source, tf_source=None, alias_source=None,
filter_havana=False, protein_coding=False, known_only=False):
# Class Objects
self.gene_list = [] # Represents gene annotation.
self.tf_list = [] # Represents TF PWM annotation.
self.alias_dict = dict() # Gene Symbol or other IDs -> ENSEMBL ID
self.symbol_dict = dict() # ENSEMBL ID -> Official gene symbol
# Initializing Required Field - Gene List
if isinstance(gene_source, list): # It can be a matrix - Used by internal methods.
self.gene_list = gene_source
if isinstance(gene_source, str): # It can be a string.
if os.path.isfile(gene_source): # The string may represent a path to a gtf file.
# FTT for TDF True
# filter_havana = False
protein_coding = False
known_only = False
self.load_gene_list(gene_source,
filter_havana=filter_havana,
protein_coding=protein_coding,
known_only=known_only)
else: # The string may represent an organism which points to a gtf file within data.config.
genome_data = GenomeData(gene_source)
self.load_gene_list(genome_data.get_annotation(),
filter_havana=filter_havana,
protein_coding=protein_coding,
known_only=known_only)
# Initializing Optional Field - TF List
if tf_source:
if isinstance(tf_source, list):
if isinstance(tf_source[0], list): # It can be a matrix
self.tf_list = tf_source
else:
mtf_file_list = []
motif_data = MotifData()
for e in tf_source:
if os.path.isfile(e): # It can be a path to a mtf file.
mtf_file_list.append(e)
else: # It can represent an organism which points to an mtf file within data.config.
mtf_file = motif_data.get_mtf_path(e)
mtf_file_list.append(mtf_file)
self.load_tf_list(mtf_file_list)
else:
pass # TODO Throw error.
# Initializing Optional Field - Alias Dictionary
if alias_source:
if isinstance(alias_source, dict): # It can be a dictionary - Used by internal methods.
self.alias_dict = alias_source
if isinstance(alias_source, str): # It can be a string.
if os.path.isfile(alias_source): # The string may represent a path to a txt alias file.
self.load_alias_dict(alias_source)
else: # The string may represent an organism which points to a txt alias file within data.config.
genome_data = GenomeData(alias_source)
self.load_alias_dict(genome_data.get_gene_alias())
else:
pass # TODO Throw error
def __init__(self, gene_source, tf_source=None, alias_source=None,
filter_havana=False, protein_coding=False, known_only=False):
# Class Objects
self.gene_list = [] # Represents gene annotation.
self.tf_list = [] # Represents TF PWM annotation.
self.alias_dict = dict() # Gene Symbol or other IDs -> ENSEMBL ID
self.symbol_dict = dict() # ENSEMBL ID -> Official gene symbol
# Initializing Required Field - Gene List
if isinstance(gene_source,list): # It can be a matrix - Used by internal methods.
self.gene_list = gene_source
if isinstance(gene_source,str): # It can be a string.
if os.path.isfile(gene_source): # The string may represent a path to a gtf file.
# FTT for TDF True
#filter_havana = False
protein_coding = False
known_only = False
self.load_gene_list(gene_source,
filter_havana=filter_havana,
protein_coding=protein_coding,
known_only=known_only)
else: # The string may represent an organism which points to a gtf file within data.config.
genome_data = GenomeData(gene_source)
self.load_gene_list(genome_data.get_annotation(),
filter_havana=filter_havana,
protein_coding=protein_coding,
known_only=known_only)
# Initializing Optional Field - TF List
if tf_source:
if isinstance(tf_source, list):
if isinstance(tf_source[0], list): # It can be a matrix
self.tf_list = tf_source
else:
mtf_file_list = []
motif_data = MotifData()
for e in tf_source:
if os.path.isfile(e): # It can be a path to a mtf file.
mtf_file_list.append(e)
else: # It can represent an organism which points to an mtf file within data.config.
mtf_file = motif_data.get_mtf_path(e)
mtf_file_list.append(mtf_file)
self.tf_list = self.load_tf_list(mtf_file_list)
else: pass # TODO Throw error.
# Initializing Optional Field - Alias Dictionary
if alias_source:
if isinstance(alias_source,dict): # It can be a dictionary - Used by internal methods.
self.alias_dict = alias_source
if isinstance(alias_source,str): # It can be a string.
if os.path.isfile(alias_source): # The string may represent a path to a txt alias file.
self.load_alias_dict(alias_source)
else: # The string may represent an organism which points to a txt alias file within data.config.
genome_data = GenomeData(alias_source)
self.load_alias_dict(genome_data.get_gene_alias())
else: pass # TODO Throw error
# gd = GenomeData(organism=genome)
# print("\t"+gd.get_annotation())
# print("\tloading GenomeData... succeeds")
genome = "hg38"
print("Checking " + genome)
annot = AnnotationSet(genome,
filter_havana=False,
protein_coding=True,
known_only=False)
# annot = AnnotationSet(genome,filter_havana=True,protein_coding=True,known_only=True)
print("\tloading AnnotationSet... succeeds")
promoters = annot.get_promoters()
print("\tPromoters " + str(len(promoters)))
gd = GenomeData(organism=genome)
print("\t" + gd.get_annotation())
print("\tloading GenomeData... succeeds")
# genome = "mm9"
# print("Checking " + genome)
# annot = AnnotationSet(genome,filter_havana=True,protein_coding=True,known_only=True)
# print("\tloading AnnotationSet... succeeds")
# promoters = annot.get_promoters()
# print("\tPromoters "+str(len(promoters)))
# gd = GenomeData(organism=genome)
# print("\t"+gd.get_annotation())
# print("\tloading GenomeData... succeeds")
# genome = "zv9"
# print("Checking " + genome)
# annot = AnnotationSet(genome,filter_havana=True,protein_coding=True,known_only=True)
# print("\tloading AnnotationSet... succeeds")
# promoters = annot.get_promoters()
# print("\tPromoters "+str(len(promoters)))
# gd = GenomeData(organism=genome)
# print("\t"+gd.get_annotation())
# print("\tloading GenomeData... succeeds")
genome = "hg38"
print("Checking " + genome)
annot = AnnotationSet(genome,filter_havana=False,protein_coding=True,known_only=False)
# annot = AnnotationSet(genome,filter_havana=True,protein_coding=True,known_only=True)
print("\tloading AnnotationSet... succeeds")
promoters = annot.get_promoters()
print("\tPromoters "+str(len(promoters)))
gd = GenomeData(organism=genome)
print("\t"+gd.get_annotation())
print("\tloading GenomeData... succeeds")
# genome = "mm9"
# print("Checking " + genome)
# annot = AnnotationSet(genome,filter_havana=True,protein_coding=True,known_only=True)
# print("\tloading AnnotationSet... succeeds")
# promoters = annot.get_promoters()
# print("\tPromoters "+str(len(promoters)))
# gd = GenomeData(organism=genome)
# print("\t"+gd.get_annotation())
# print("\tloading GenomeData... succeeds")
# genome = "zv9"
# print("Checking " + genome)
# annot = AnnotationSet(genome,filter_havana=True,protein_coding=True,known_only=True)
