EpiTOC3#
Index#
Let’s first import some packages:
[1]:
import os
import inspect
import shutil
import json
import torch
import pandas as pd
import pyaging as pya
Instantiate model class#
[2]:
def print_entire_class(cls):
source = inspect.getsource(cls)
print(source)
print_entire_class(pya.models.EpiTOC3)
class EpiTOC3(epiTOC2):
pass
[3]:
model = pya.models.EpiTOC3()
Define clock metadata#
[4]:
model.metadata["clock_name"] = "epitoc3"
model.metadata["data_type"] = "DNA methylation" # Paper: The study constructs clocks from DNA methylation measurements.
model.metadata["species"] = "Homo sapiens" # Paper: The analyzed samples and clock are human.
model.metadata["year"] = 2020
model.metadata["approved_by_author"] = "⌛"
model.metadata["citation"] = "Teschendorff, Andrew E. \"A comparison of epigenetic mitotic-like clocks for cancer risk prediction.\" Genome Medicine 12 (2020): 56."
model.metadata["doi"] = "https://doi.org/10.1186/s13073-020-00752-3"
model.metadata["notes"] = "Code-defined 170-CpG extension of the dynamic mitotic model. Official EpiMitClocks data show that all 170 sites are a subset of the 371 stemTOC vivo-mitCpGs derived from fetal/neonatal references, six normal proliferating cell lines, and three adult whole-blood cohorts. The assigned 2020 dynamic-model paper does not name or define epiTOC3."
model.metadata["research_only"] = None
model.metadata["tissue"] = ["cultured primary human cells", "whole blood", "multi-tissue", "cord blood"] # Paper: epiTOC3 uses 170 unmethylated population-doubling-associated CpGs.
model.metadata["predicts"] = ["mitotic age"] # Paper: The function directly estimates cumulative stem-cell divisions in a tissue.
model.metadata["training_target"] = ["population doublings"] # Paper: The epiTOC3 signature comprises population-doubling-associated CpGs.
model.metadata["unit"] = ["cell divisions per stem cell"] # Paper: The returned TNSC is cumulative stem-cell divisions.
model.metadata["model_type"] = "dynamic methylation transmission model" # Paper: The function describes epiTOC3 as building on a dynamic DNA methylation-gain model.
model.metadata["platform"] = ["Illumina 450K", "Illumina EPIC"] # Paper: Input is an Illumina 450k/EPIC DNAm beta matrix.
model.metadata["population"] = "all ages" # Paper: The derivation uses fetal/neonatal reference tissues, six normal cell lines, and adult whole-blood cohorts.
model.metadata["journal"] = "Genome Medicine"
model.metadata["last_author"] = "Andrew E. Teschendorff"
model.metadata["n_features"] = 170
model.metadata["citations"] = 155
model.metadata["citations_date"] = "2026-07-05"
Download clock dependencies#
[5]:
supplementary_url = "https://raw.githubusercontent.com/Duzhaozhen/OmniAge/c10fbe8cb92957520fbff1d55ae1def0691252e5/OmniAgePy/src/omniage/data/epiTOC3.csv"
supplementary_file_name = "coefficients.csv"
os.system(f"curl -sL -o {supplementary_file_name} {supplementary_url}")
[5]:
0
Load features#
[6]:
df = pd.read_csv('coefficients.csv')
model.features = df['probe'].astype(str).tolist()
Load weights into base model#
[7]:
model.delta = torch.tensor(df['delta'].values, dtype=torch.float32).unsqueeze(0)
model.beta0 = torch.tensor(df['beta0'].values, dtype=torch.float32).unsqueeze(0)
[8]:
model.base_model = None
Load reference values#
[9]:
model.reference_values = [-1]*len(model.features)
Load preprocess and postprocess objects#
[10]:
model.preprocess_name = "nan_to_zero"
model.preprocess_dependencies = None
[11]:
model.postprocess_name = None
model.postprocess_dependencies = None
Check all clock parameters#
[12]:
pya.utils.print_model_details(model)
%==================================== Model Details ====================================%
Model Attributes:
training: True
metadata: {'approved_by_author': '⌛',
'citation': 'Teschendorff, Andrew E. "A comparison of epigenetic mitotic-like '
'clocks for cancer risk prediction." Genome Medicine 12.1 (2020): '
'56.',
'clock_name': 'epitoc3',
'data_type': 'methylation',
'doi': 'https://doi.org/10.1186/s13073-020-00752-3',
'notes': None,
'research_only': None,
'species': 'Homo sapiens',
'version': None,
'year': 2020}
reference_values: [-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1]... [Total elements: 170]
preprocess_name: 'nan_to_zero'
preprocess_dependencies: None
postprocess_name: None
postprocess_dependencies: None
features: ['cg23062112', 'cg18560328', 'cg21229268', 'cg11250773', 'cg12892303', 'cg12781700', 'cg12065366', 'cg17712694', 'cg12799781', 'cg01718742', 'cg10263370', 'cg16491617', 'cg11832210', 'cg07686635', 'cg26510017', 'cg15745900', 'cg15402399', 'cg05009934', 'cg10094616', 'cg06010588', 'cg23847712', 'cg17370163', 'cg13428480', 'cg22620221', 'cg16437728', 'cg25984671', 'cg07882671', 'cg04342955', 'cg14482313', 'cg14537533']... [Total elements: 170]
base_model_features: None
base_model: None
delta: [9.999999747378752e-05, 9.999999747378752e-05, 9.999999747378752e-05, 0.0002500000118743628, 9.999999747378752e-05, 9.999999747378752e-05, 9.999999747378752e-05, 9.999999747378752e-05, 4.999999873689376e-05, 0.0002500000118743628, 0.0002500000118743628, 9.999999747378752e-05, 9.999999747378752e-06, 9.999999747378752e-06, 4.999999873689376e-05, 4.999999873689376e-05, 9.999999747378752e-06, 4.999999873689376e-05, 4.999999873689376e-05, 0.0002500000118743628, 0.0002500000118743628, 0.0002500000118743628, 4.999999873689376e-05, 0.0002500000118743628, 9.999999747378752e-05, 9.999999747378752e-05, 9.999999747378752e-05, 4.999999873689376e-05, 0.0002500000118743628, 9.999999747378752e-05]... [Tensor of shape torch.Size([1, 170])]
beta0: [0.05000000074505806, 0.019999999552965164, 0.029999999329447746, 0.0, 0.03999999910593033, 0.03999999910593033, 0.05000000074505806, 0.05000000074505806, 0.029999999329447746, 0.009999999776482582, 0.019999999552965164, 0.029999999329447746, 0.03999999910593033, 0.05000000074505806, 0.019999999552965164, 0.03999999910593033, 0.05000000074505806, 0.03999999910593033, 0.05000000074505806, 0.05000000074505806, 0.05000000074505806, 0.05000000074505806, 0.03999999910593033, 0.05000000074505806, 0.05000000074505806, 0.05000000074505806, 0.019999999552965164, 0.05000000074505806, 0.0, 0.019999999552965164]... [Tensor of shape torch.Size([1, 170])]
%==================================== Model Details ====================================%
Model Structure:
%==================================== Model Details ====================================%
Model Parameters and Weights:
%==================================== Model Details ====================================%
Basic test#
[13]:
torch.manual_seed(42)
input = torch.randn(10, len(model.features), dtype=float)
model.eval()
model.to(float)
pred = model(input)
pred
[13]:
tensor([[ 1921.6122],
[-4698.6198],
[ 1365.4274],
[-4247.0676],
[ -35.5126],
[ 116.7216],
[ 7356.7134],
[-3367.8507],
[ 7780.6034],
[-7170.8200]], dtype=torch.float64)
Save torch model#
[14]:
torch.save(model, f"../weights/{model.metadata['clock_name']}.pt")
Clear directory#
[15]:
# Function to remove a folder and all its contents
def remove_folder(path):
try:
shutil.rmtree(path)
print(f"Deleted folder: {path}")
except Exception as e:
print(f"Error deleting folder {path}: {e}")
# Get a list of all files and folders in the current directory
all_items = os.listdir('.')
# Loop through the items
for item in all_items:
# Check if it's a file and does not end with .ipynb
if os.path.isfile(item) and not item.endswith('.ipynb'):
os.remove(item)
print(f"Deleted file: {item}")
# Check if it's a folder
elif os.path.isdir(item):
remove_folder(item)
Deleted file: coefficients.csv