DownSyndrome#

Index#

Instantiate model class
Define clock metadata
Download clock dependencies
Load features
Load weights into base model
Load reference values
Load preprocess and postprocess objects
Check all clock parameters
Basic test
Save torch model
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.DownSyndrome)

class DownSyndrome(LinearReferenceClock):
    pass

[3]:

model = pya.models.DownSyndrome()

Define clock metadata#

[4]:

model.metadata["clock_name"] = "downsyndrome"
model.metadata["data_type"] = "DNA methylation"  # Paper: genome-wide DNA methylation data
model.metadata["species"] = "Homo sapiens"  # Paper: 196 DS and 439 non-DS newborns
model.metadata["year"] = 2021
model.metadata["approved_by_author"] = "⌛"
model.metadata["citation"] = "Muskens, Ivo S., et al. \"The genome-wide impact of trisomy 21 on DNA methylation and its implications for hematopoiesis.\" Nature Communications 12 (2021): 821."
model.metadata["doi"] = "https://doi.org/10.1038/s41467-021-21064-z"
model.metadata["notes"] = "Implementation-derived Down-syndrome-associated methylation projection score: pyaging computes a zero-intercept weighted sum of 652 neonatal blood-spot beta values using the paper's autosomal EWAS beta_overall effect estimates. The paper presents an EWAS, not a trained or validated Down syndrome classifier."
model.metadata["research_only"] = None
model.metadata["tissue"] = ["neonatal blood spots"]  # Paper: newborn blood samples
model.metadata["predicts"] = ["Down syndrome methylation score"]  # Paper: The output is the sum of methylation beta values weighted by beta_overall EWAS estimates.
model.metadata["training_target"] = ["not applicable"]  # Paper: The EWAS compares 196 newborns with Down syndrome and 439 without Down syndrome.
model.metadata["unit"] = ["unitless"]  # Paper: Methylation beta values and beta_overall weights are summed without calibration.
model.metadata["model_type"] = "weighted linear score"  # Paper: weights = beta_overall; intercept = 0.0
model.metadata["platform"] = ["Illumina EPIC"]  # Paper: Illumina Infinium MethylationEPIC Beadchip
model.metadata["population"] = "newborns"  # Paper: 196 DS and 439 non-DS newborns
model.metadata["journal"] = "Nature Communications"
model.metadata["last_author"] = "Adam J. de Smith"
model.metadata["n_features"] = 652
model.metadata["citations"] = 62
model.metadata["citations_date"] = "2026-07-05"

Download clock dependencies#

[5]:

os.system(f"curl -sL -o downsyndrome.xlsx https://static-content.springer.com/esm/art%3A10.1038%2Fs41467-021-21064-z/MediaObjects/41467_2021_21064_MOESM6_ESM.xlsx")

[5]:

Load features#

[6]:

df = pd.read_excel('downsyndrome.xlsx', sheet_name='EWAS_autosomes', skiprows=2)
model.features = df['CpG'].tolist()

/Users/lucascamillo/pyaging/.venv/lib/python3.13/site-packages/openpyxl/worksheet/_reader.py:329: UserWarning: Unknown extension is not supported and will be removed
  warn(msg)

Load weights into base model#

[7]:

weights = torch.tensor(df['beta_overall'].tolist()).unsqueeze(0).float()
intercept = torch.tensor([0.0]).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 = 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': 'Muskens, Ivo S., et al. "Germline aberrant methylation '
             'associated with Down syndrome and its impact on childhood acute '
             'lymphoblastic leukemia risk." Nature Communications 12.1 (2021): '
             '821.',
 'clock_name': 'downsyndrome',
 'data_type': 'methylation',
 'doi': 'https://doi.org/10.1038/s41467-021-21064-z',
 'notes': None,
 'research_only': None,
 'species': 'Homo sapiens',
 'version': None,
 'year': 2021}
reference_values: None
preprocess_name: None
preprocess_dependencies: None
postprocess_name: None
postprocess_dependencies: None
features: ['cg07741821', 'cg02993069', 'cg12477880', 'cg08882472', 'cg24942416', 'cg07841633', 'cg00994804', 'cg11218872', 'cg02451831', 'cg13382072', 'cg24020235', 'cg17239923', 'cg19030331', 'cg03142697', 'cg24999883', 'cg12679760', 'cg23719650', 'cg11972401', 'cg23565347', 'cg19765472', 'cg04599341', 'cg16831070', 'cg06758350', 'cg11151981', 'cg08620751', 'cg08849601', 'cg02680932', 'cg22539182', 'cg11075561', 'cg04131721']... [Total elements: 652]
base_model_features: None

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

base_model: LinearModel(
  (linear): Linear(in_features=652, out_features=1, bias=True)
)

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

base_model.linear.weight: [-0.28867992758750916, 0.19754931330680847, 0.38263070583343506, 0.15737323462963104, -0.17506907880306244, -0.3285122215747833, 0.38840898871421814, -0.12085756659507751, -0.15854349732398987, 0.179888516664505, 0.18759849667549133, 0.23218844830989838, 0.16510215401649475, 0.3955685794353485, -0.0671553760766983, 0.18286938965320587, -0.10707408934831619, 0.05787608027458191, -0.09203432500362396, 0.2067182958126068, 0.19878044724464417, -0.061179906129837036, 0.28260132670402527, -0.062310606241226196, 0.1342443823814392, -0.07032307982444763, -0.07384317368268967, -0.14031463861465454, -0.22928495705127716, -0.0856233760714531]... [Tensor of shape torch.Size([1, 652])]
base_model.linear.bias: tensor([0.])

%==================================== 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([[-4.8313],
        [-1.6036],
        [ 4.0192],
        [-1.2427],
        [ 1.9975],
        [-1.7729],
        [ 0.8015],
        [-1.8500],
        [ 1.6546],
        [ 0.3752]], dtype=torch.float64, grad_fn=<AddmmBackward0>)

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: downsyndrome.xlsx