def __init__(
self,
gtf_file,
fasta_file,
side='5prime', # 5prime/3prime
target_file=None,
MISO_AS=False,
label_col='event_name'):
self.genes = loadgene(gtf_file)
self.fasta_file = fasta_file
self.fasta = None # open the file later
if side in ["5prime", "3prime"]:
self.side = side
else:
raise Exception("side should be 5prime or 3prime")
if self.side == "5prime":
self.overhang_l = 3
self.overhang_r = 6
else:
self.overhang_l = 3
self.overhang_r = 20
if target_file is not None:
self.Y = Target(target_file, label_col)
else:
self.Y = None
self.MISO_AS = MISO_AS
if not MISO_AS:
self.spliceSites = self.get_spliceSites()
self._name = None
self._species = None
def __init__(self,
gtf_file,
fasta_file,
overhang=80,
MISO_AS=False): # intron + ~ bp exon from both side
self.genes = loadgene(gtf_file)
self.fasta = FastaFile(fasta_file)
self.overhang = overhang
self.MISO_AS = MISO_AS
if not MISO_AS:
self.spliceSites = self.get_spliceSites()
self._name = None
self._species = None
def __init__(self, gtf_file, fasta_file, length=70):
self.genes = loadgene(gtf_file)
self.fasta = FastaFile(fasta_file)
self.length = length
self.branches = self.get_branches()