def __init__(self, pwm_list=[], stddev=0.05, seed=None):
if len(pwm_list) > 0 and isinstance(pwm_list[0], dict):
pwm_list = [PWM.from_config(pwm) for pwm in pwm_list]
self.stddev = stddev
self.seed = seed
self.pwm_list = pwm_list
_check_pwm_list(pwm_list)
def __init__(self, pwm_list=[], kernel_size=None, mean_max_scale=0.):
# handle pwm_list as a dictionary
if len(pwm_list) > 0 and isinstance(pwm_list[0], dict):
pwm_list = [PWM.from_config(pwm) for pwm in pwm_list]
if kernel_size is None:
kernel_size = len(pwm_list)
self.pwm_list = pwm_list
self.kernel_size = kernel_size
self.mean_max_scale = mean_max_scale
_check_pwm_list(pwm_list)
def __init__(self,
pwm_list=[],
stddev=0.05,
seed=None,
background_probs=DEFAULT_BASE_BACKGROUND,
add_noise_before_Pwm2Pssm=True):
if len(pwm_list) > 0 and isinstance(pwm_list[0], dict):
pwm_list = [PWM.from_config(pwm) for pwm in pwm_list]
self.pwm_list = pwm_list
_check_pwm_list(pwm_list)
self.stddev = stddev
self.seed = seed
self.background_probs = background_probs
self.add_noise_before_Pwm2Pssm = add_noise_before_Pwm2Pssm
def __init__(self,
pwm_list=[],
kernel_size=None,
mean_max_scale=0.,
background_probs=DEFAULT_BASE_BACKGROUND):
# handle pwm_list as a dictionary
if len(pwm_list) > 0 and isinstance(pwm_list[0], dict):
pwm_list = [PWM.from_config(pwm) for pwm in pwm_list]
if kernel_size is None:
kernel_size = len(pwm_list)
_check_pwm_list(pwm_list)
self.pwm_list = pwm_list
self.kernel_size = kernel_size
self.mean_max_scale = mean_max_scale
self.background_probs = background_probs
def get_branchpoint_pwm_list(cache=True):
l = []
if os.path.isfile(BR_PWM) and cache:
l.append(PWM.from_config(read_json(BR_PWM)))
else:
dt = pd.read_csv(
DATA_ROOT +
"/Splice_branchpoints/processed/branchpointer/train/filteredDescr.csv"
)
# colmeans
dt = dt[dt.set == "HC"]
dtseq = dt[dt.columns[dt.columns.str.match("^seq_")]] - 1
pwm = np.array(dtseq.mean()).reshape((-1, 4))
assert np.allclose(pwm.sum(1), 1)
p = PWM(pwm, name="U2_branchpoint")
write_json(p.get_config(), BR_PWM)
l.append(p)
l.append(
PWM(0.05 + np.loadtxt(BR_SPLICE_RACK_PATH + "/GT_AG_U12.txt"),
"GT_AG_U12_branchpoint"))
return l