'C': 1.0,
'mu': 1.0,
'alpha': 1.0,
'h**o': 27.2114,
'lumo': 27.2114,
'gap': 27.2114,
'r2': 1.0,
'zpve': 27211.4,
'u0': 27.2114,
'u298': 27.2114,
'h298': 27.2114,
'g298': 27.2114,
'cv': 1.0,
'u0_atom': 27.2114,
'u298_atom': 27.2114,
'h298_atom': 27.2114,
'g298_atom': 27.2114,
'cv_atom': 1.0
}
for label in labels:
cf = conversion[label]
rloader = res.ResultsENN('smp-' + label, reps=[1, 2, 3, 4, 5])
results = rloader.get_all_predictions()
summary = met.evaluate_average(results,
metric=met.mae,
verbose=True,
select=3)
summary = [(cf * s[0], cf * s[1]) for s in summary]
print('%9s: %6.3f \pm %6.3f' % (label, *summary[2]))
'A': 1.0,
'B': 1.0,
'C': 1.0,
'mu': 1.0,
'alpha': 1.0,
'h**o': 27.2114,
'lumo': 27.2114,
'gap': 27.2114,
'r2': 1.0,
'zpve': 27211.4,
'u0': 27.2114,
'u298': 27.2114,
'h298': 27.2114,
'g298': 27.2114,
'cv': 1.0,
'u0_atom': 27.2114,
'u298_atom': 27.2114,
'h298_atom': 27.2114,
'g298_atom': 27.2114,
'cv_atom': 1.0
}
for label in labels:
name = f'logs/smp_test_{label}/smp'
cf = conversion[label]
rloader = res.ResultsGNN(name, reps=[0, 1, 2])
results = rloader.get_all_predictions()
summary = met.evaluate_average(results, metric=met.mae, verbose=False)
summary = [(cf * s[0], cf * s[1]) for s in summary]
print('%9s: %6.3f \pm %6.3f' % (label, *summary[2]))
import numpy as np
import torch
import atom3d.util.results as res
import atom3d.util.metrics as met
# Define the training run
name = 'logs/msp_test/msp'
print(name)
# Load training results
rloader = res.ResultsGNN(name, reps=[0, 1, 2])
results = rloader.get_all_predictions()
# Calculate and print results
summary = met.evaluate_average(results, metric=met.auroc, verbose=False)
print('Test AUROC: %6.3f \pm %6.3f' % summary[2])
summary = met.evaluate_average(results, metric=met.auprc, verbose=False)
print('Test AUPRC: %6.3f \pm %6.3f' % summary[2])
import sys
import numpy as np
import torch
import atom3d.util.results as res
import atom3d.util.metrics as met
seqid = sys.argv[1]
# Define the training run
name = f'logs/lba_test_{seqid}/lba'
print(name)
# Load training results
rloader = res.ResultsGNN(name, reps=[0, 1, 2])
results = rloader.get_all_predictions()
# Calculate and print results
summary = met.evaluate_average(results, metric=met.rmse, verbose=False)
print('Test RMSE: %6.3f \pm %6.3f' % summary[2])
summary = met.evaluate_average(results, metric=met.spearman, verbose=False)
print('Test Spearman: %6.3f \pm %6.3f' % summary[2])
summary = met.evaluate_average(results, metric=met.pearson, verbose=False)
print('Test Pearson: %6.3f \pm %6.3f' % summary[2])
# Load training results
rloader = res.ResultsGNN(name, reps=[0, 1, 2])
results = rloader.get_target_specific_predictions()
# Calculate and print results
summary = met.evaluate_per_target_average(results['per_target'],
metric=met.spearman,
verbose=False)
print('Test Spearman (per-target): %6.3f \pm %6.3f' % summary[2])
summary = met.evaluate_per_target_average(results['per_target'],
metric=met.pearson,
verbose=False)
print('Test Pearson (per-target): %6.3f \pm %6.3f' % summary[2])
summary = met.evaluate_per_target_average(results['per_target'],
metric=met.kendall,
verbose=False)
print('Test Kendall (per-target): %6.3f \pm %6.3f' % summary[2])
summary = met.evaluate_average(results['global'],
metric=met.spearman,
verbose=False)
print('Test Spearman (global): %6.3f \pm %6.3f' % summary[2])
summary = met.evaluate_average(results['global'],
metric=met.pearson,
verbose=False)
print('Test Pearson (global): %6.3f \pm %6.3f' % summary[2])
summary = met.evaluate_average(results['global'],
metric=met.kendall,
verbose=False)
print('Test Kendall (global): %6.3f \pm %6.3f' % summary[2])