'''Script to execute training. Assumes projects have already been created
and tiles have already been extracted from slides.
'''
import click
import multiprocessing
import re
from typing import List
from biscuit.experiment import Experiment, ALL_EXP
# -----------------------------------------------------------------------------
def num_range(s: str) -> List[int]:
'''Accept either a comma separated list of numbers 'a,b,c' or a
range 'a-c' and return as a list of ints.
'''
range_re = re.compile(r'^(\d+)-(\d+)$')
m = range_re.match(s)
if m:
return list(range(int(m.group(1)), int(m.group(2))+1))
vals = s.split(',')
return [int(x) for x in vals]
# -----------------------------------------------------------------------------
@click.command()
@click.option('--train_project', default='projects/training', type=str, help='Override training project')
@click.option('--eval_project', default='projects/evaluation',type=str, help='Override eval project')
@click.option('--outcome', type=str, help='Outcome (annotation header) that assigns class labels.', default='cohort', show_default=True)
@click.option('--outcome1', type=str, help='First class label.', default='LUAD', show_default=True)
@click.option('--outcome2', type=str, help='Second class label.', default='LUSC', show_default=True)
@click.option('--steps', type=num_range, help='Training steps to perform')
@click.option('--reg', type=bool, help='Train regular models', default=True)
@click.option('--ratio', type=bool, help='Train ratio models', default=True)
@click.option('--gan', type=bool, help='Train gan models', default=False)
def train_models(
train_project,
eval_project,
outcome,
outcome1,
outcome2,
steps=None,
reg=True,
ratio=True,
gan=False,
):
# --- Configure experiments -----------------------------------------------
experiment = Experiment(
train_project,
eval_projects=[eval_project],
outcome=outcome,
outcome1=outcome1,
outcome2=outcome2,
outdir='results')
if steps is None:
steps = range(7)
to_run = []
# Configure regular experiments
if reg:
reg1 = experiment.config('{}', ALL_EXP, 1, order='f')
reg2 = experiment.config('{}2', ALL_EXP, 1, order='f', order_col='order2')
rev1 = experiment.config('{}_R', ALL_EXP, 1, order='r')
rev2 = experiment.config('{}_R2', ALL_EXP, 1, order='r', order_col='order2')
to_run += [reg1, reg2, rev1, rev2]
# Configure 3:1 and 10:1 ratio experiments
if ratio:
ratio_exp = list('AMDPGZ')
ratio_3 = experiment.config('{}_3', ratio_exp, 3, order='f')
ratio_3_rev = experiment.config('{}_R_3', ratio_exp, 3, order='r')
ratio_10 = experiment.config('{}_10', ratio_exp, 10, order='f')
ratio_10_rev = experiment.config('{}_R_10', ratio_exp, 10, order='r')
to_run += [ratio_3, ratio_3_rev, ratio_10, ratio_10_rev]
# GAN experiments
if gan:
_g = list('RALMNDOPQGWY') + ['ZA', 'ZC']
gan_exp = {}
gan_exp.update(experiment.config('{}_g10', _g, 1, gan=0.1, order='f'))
gan_exp.update(experiment.config('{}_R_g10', _g, 1, gan=0.1, order='r'))
gan_exp.update(experiment.config('{}_g20', _g, 1, gan=0.2, order='f'))
gan_exp.update(experiment.config('{}_R_g20', _g, 1, gan=0.2, order='r'))
gan_exp.update(experiment.config('{}_g30', _g, 1, gan=0.3, order='f'))
gan_exp.update(experiment.config('{}_R_g30', _g, 1, gan=0.3, order='r'))
gan_exp.update(experiment.config('{}_g40', _g, 1, gan=0.4, order='f'))
gan_exp.update(experiment.config('{}_R_g40', _g, 1, gan=0.4, order='r'))
gan_exp.update(experiment.config('{}_g50', _g, 1, gan=0.5, order='f'))
gan_exp.update(experiment.config('{}_R_g50', _g, 1, gan=0.5, order='r'))
to_run += [gan_exp]
# --- Train experiments ---------------------------------------------------
for exp in to_run:
experiment.run(exp, steps=steps)
if __name__ == '__main__':
multiprocessing.freeze_support()
train_models() # pylint: disable=no-value-for-parameter
Datasets
Models
