Bohlin#

Index#

  1. Instantiate model class

  2. Define clock metadata

  3. Download clock dependencies

  4. Load features

  5. Load weights into base model

  6. Load reference values

  7. Load preprocess and postprocess objects

  8. Check all clock parameters

  9. Basic test

  10. Save torch model

  11. Clear directory

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.Bohlin)

class Bohlin(LinearReferenceClock):
    def postprocess(self, x):
        """Model returns gestational age in days; convert to weeks."""
        return x / 7.0


[3]:

model = pya.models.Bohlin()

Define clock metadata#

[4]:

model.metadata["clock_name"] = "bohlin"
model.metadata["data_type"] = "DNA methylation"  # Paper: The predictor uses cord-blood DNA methylation beta values.
model.metadata["species"] = "Homo sapiens"  # Paper: The study analyzed newborns from the human MoBa cohort.
model.metadata["year"] = 2016
model.metadata["approved_by_author"] = "⌛"
model.metadata["citation"] = "Bohlin, J., Håberg, S. E., Magnus, P. et al. Prediction of gestational age based on genome-wide differentially methylated regions. Genome Biology 17, 207 (2016)."
model.metadata["doi"] = "https://doi.org/10.1186/s13059-016-1063-4"
model.metadata["notes"] = "Official minimum-lambda variant of the Bohlin gestational-age LASSO: pyaging implements the 251-CpG lambda.min model and converts its day-scale output to weeks. The paper/package default one-standard-error variant uses 96 CpGs and has nearly identical predictive performance."
model.metadata["research_only"] = None
model.metadata["tissue"] = ["cord blood"]  # Paper: Training used Illumina 450K arrays based on newborn cord-blood DNA.
model.metadata["predicts"] = ["gestational age"]  # Paper: The model estimates gestational age after birth from cord-blood methylation.
model.metadata["training_target"] = ["gestational age"]  # Paper: The primary, better-performing model was trained against ultrasound gestational age.
model.metadata["unit"] = ["weeks"]  # Paper: The coefficient formula is in days and the pyaging implementation divides the result by seven.
model.metadata["model_type"] = "LASSO regression"  # Paper: Gestational-age prediction used glmnet LASSO regression.
model.metadata["platform"] = ["Illumina 450K"]  # Paper: MoBa training methylomes were measured on the Illumina HumanMethylation450 platform.
model.metadata["population"] = "newborns"  # Paper: The model was trained in 1,068 MoBa newborns and tested in 685 additional MoBa newborns.
model.metadata["journal"] = "Genome Biology"
model.metadata["last_author"] = "Wenche Nystad"
model.metadata["n_features"] = 251
model.metadata["citations"] = 237
model.metadata["citations_date"] = "2026-07-05"

Download clock dependencies#

[5]:

os.system(f"curl -sL -o coefficients.csv https://raw.githubusercontent.com/Duzhaozhen/OmniAge/c10fbe8cb92957520fbff1d55ae1def0691252e5/OmniAgePy/src/omniage/data/Bohlin_GA.csv")

[5]:

0

Load features#

[6]:

df = pd.read_csv('coefficients.csv')
mask = df['probe'].astype(str).str.lower().isin(['intercept', '(intercept)'])
intercept_value = float(df.loc[mask, 'coef'].iloc[0]) if mask.any() else 0.0
coef_df = df.loc[~mask].reset_index(drop=True)
model.features = coef_df['probe'].tolist()

Load weights into base model#

[7]:

weights = torch.tensor(coef_df['coef'].tolist()).unsqueeze(0).float()
intercept = torch.tensor([intercept_value]).float()

[8]:

base_model = pya.models.LinearModel(input_dim=len(model.features))

base_model.linear.weight.data = weights.float()
base_model.linear.bias.data = intercept.float()

model.base_model = base_model

Load reference values#

[9]:

model.reference_values = None

Load preprocess and postprocess objects#

[10]:

model.preprocess_name = None
model.preprocess_dependencies = None

[11]:

model.postprocess_name = 'days_to_weeks'
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': 'Bohlin, Jon, et al. "Prediction of gestational age based on '
             'genome-wide differentially methylated regions." Genome Biology '
             '17.1 (2016): 207.',
 'clock_name': 'bohlin',
 'data_type': 'methylation',
 'doi': 'https://doi.org/10.1186/s13059-016-1063-4',
 'notes': None,
 'research_only': None,
 'species': 'Homo sapiens',
 'version': None,
 'year': 2016}
reference_values: None
preprocess_name: None
preprocess_dependencies: None
postprocess_name: 'days_to_weeks'
postprocess_dependencies: None
features: ['cg00099441', 'cg00117869', 'cg00303541', 'cg00602416', 'cg00711496', 'cg00898111', 'cg01091356', 'cg01139861', 'cg01190109', 'cg01281797', 'cg01359999', 'cg01382072', 'cg01470456', 'cg01505275', 'cg01635555', 'cg01697487', 'cg01749742', 'cg01833485', 'cg01847441', 'cg02324006', 'cg02358664', 'cg02378208', 'cg02405476', 'cg02567958', 'cg02576394', 'cg02642822', 'cg02779037', 'cg02853322', 'cg02985694', 'cg03098721']... [Total elements: 251]
base_model_features: None

%==================================== Model Details ====================================%
Model Structure:

base_model: LinearModel(
  (linear): Linear(in_features=251, out_features=1, bias=True)
)

%==================================== Model Details ====================================%
Model Parameters and Weights:

base_model.linear.weight: [1.679152488708496, -2.247448444366455, -8.844316482543945, 3.3127026557922363, -1.2231009006500244, 0.2817184627056122, 0.6226192712783813, 1.777445673942566, 9.944578170776367, 25.59168243408203, 4.947725772857666, 3.4168262481689453, -0.4994896650314331, -2.2033567428588867, 5.669358730316162, 1.2348579168319702, 0.014705928973853588, 3.155426263809204, 2.1350390911102295, -12.360982894897461, 1.5933818817138672, -2.458742618560791, 11.212024688720703, -16.415714263916016, -4.602630138397217, -3.348295211791992, -2.3952744007110596, 7.600930213928223, -4.37376594543457, 8.569387435913086]... [Tensor of shape torch.Size([1, 251])]
base_model.linear.bias: tensor([277.2421])

%==================================== 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([[67.3303],
        [36.4561],
        [18.0902],
        [37.1463],
        [50.0283],
        [48.1619],
        [16.2032],
        [38.6205],
        [60.7712],
        [54.3381]], dtype=torch.float64, grad_fn=<DivBackward0>)

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