OSCR

Large-scale synthetic data enable digital twins of human excitable cells.

Code ↔ Paper

6 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 6 matches
  1. [1] § Methods › Training and loss function ↔ network/iPSC_training/test21/DL1.py, lines 135–197 · score 0.66 · Keras, monitored, metrics, absolute, batches, epoch
  2. [2] § Methods › Training and loss function ↔ network/iPSC_training/test21/DL1.py, lines 135–197 · score 0.59 · squared error, Adam, loss, metric, optimizer, trained
  3. [3] § Methods › iPSC-CM electrophysiological model ↔ Fig5_experiments_vs_model/simulated/fig5A_sim/ipsc_function.h, lines 18–67 · score 0.54 · Kernik Clancy, CM model, variables, iPSC, gating, IKs
  4. [4] § Methods › iPSC-CM electrophysiological model ↔ Fig5_experiments_vs_model/simulated/fig5B_sim/ipsc_function.h, lines 18–67 · score 0.54 · Kernik Clancy, CM model, variables, iPSC, gating, IKs
  5. [5] § Results › Deep learning approaches for voltage clamp optimization and parameter estimation ↔ network/iPSC_training/test21/DL1.py, lines 200–254 · score 0.53 · absolute error, error distribution, epochs, predictive, training, network
  6. [6] § Results › Deep learning approaches for voltage clamp optimization and parameter estimation ↔ network/iPSC_training/ipsc_function.h, lines 1–17 · score 0.51 · induced pluripotent stem, cell derived cardiomyocytes, model parameters, iPSC, network, training

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

Python · 254 lines · 8 KB · CC-BY-NC-4.0 · 3 matches

  1. #Created by Mao-Tsuen Jeng
  2. #Colleen Clancy Lab @ UC davis
  3. import tensorflow as tf
  4. import numpy as np
  5. import tensorflow.keras as keras
  6. from matplotlib import pyplot as plt
  7. import pickle
  8. import os
  9. import glob
  11. #import csv
  12. from tensorflow.keras.models import Sequential, Model
  13. #from keras.layers.core import Dense, Dropout, Activation
  14. from tensorflow.keras.layers import Dense, Dropout, Activation, LSTM, SimpleRNN
  15. from tensorflow.keras.optimizers import SGD, Adagrad
  16. from tensorflow.keras import layers
  17. from tensorflow.keras.optimizers import RMSprop
  18. from tensorflow.keras.callbacks import EarlyStopping
  19. from tensorflow.keras.utils import plot_model
  21. from tensorflow.python.client import device_lib
  22. print(device_lib.list_local_devices())
  24. #from tensorflow.keras import backend as K
  25. #K.tensorflow_backend._get_available_gpus()
  29. idxs = np.arange(0, 7625, 1 )
  33. l1 = np.size(idxs)
  35. lt = l1
  36. # This determines the number of inputs
  37. #x = np.zeros((n, lt))
  38. x1= np.zeros((1, lt))
  39. # This determines the number of outputs
  40. sy = 52
  41. #y = np.zeros((n, sy))
  42. y1 = np.zeros((1, sy))
  44. foldername = '../valid'
  45. filename = foldername + '/itotal_cell1' + '.txt'
  46. raw_data = open( filename, 'rt' )
  47. data = np.loadtxt( raw_data, delimiter='\t')
  48. raw_data.close()
  49. id1 = 0
  50. x1[id1, 0:l1] = data[idxs]
  54. filename3 = foldername + '/parameters_cell1' + '.txt'
  55. raw_data = open( filename3, 'rt' )
  56. data3 = np.loadtxt( raw_data, delimiter='\t')
  57. raw_data.close()
  58. y1[id1,:] = data3[:]
  62. x = np.load('../itotal_all.npy')
  63. y = np.load('../parameters_all.npy')
  64. n = x.shape[0]
  68. n1 = 0
  69. id2 = 0
  72. tf.keras.backend.set_floatx('float64')
  74. nk = 1024 # sy * 16
  76. model = Sequential()
  77. model.add(Dense(nk, input_shape=(lt, ), activation='tanh', dtype=tf.float64))
  78. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  79. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  80. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  81. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  82. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  83. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  84. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  85. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  86. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  87. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  88. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  89. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  90. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  91. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  92. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  93. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  94. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  95. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  96. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  97. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  98. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  99. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  100. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  101. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  102. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  103. model.add(Dense(nk, activation='tanh', dtype=tf.float64))
  104. model.add(Dense(sy, activation='linear', dtype=tf.float64))
  107. batch_size = np.round(n*0.01) #10000
  108. n1= np.round(n*0.1).astype(int) #100000
  109. input_size = n - 3 * n1
  112. x_tr = tf.cast(x[0:n-n1*3,], tf.float64)
  115. y_tr = tf.cast(y[0:n-n1*3,], tf.float64)
  118. x_val = tf.cast(x[n-3*n1:n-2*n1,], tf.float64)
  120. y_val = tf.cast(y[n-3*n1:n-2*n1,], tf.float64)
  122. x_test = tf.cast(x[n-2*n1:n,], tf.float64)
  124. y_test = tf.cast(y[n-2*n1:n,], tf.float64)
  127. train_dataset = tf.data.Dataset.from_tensor_slices((x_tr, y_tr))
  128. train_dataset = train_dataset.shuffle(input_size)
  129. train_dataset = train_dataset.batch(batch_size).cache()
  130. train_dataset = train_dataset.prefetch(tf.data.experimental.AUTOTUNE)
  133. model_wrapper = model
  135. initial_learning_rate = 0.0001
  136. lr_schedule = keras.optimizers.schedules.ExponentialDecay(
  137. initial_learning_rate,
  138. decay_steps=100,
  139. decay_rate=0.99,
  140. staircase=True)
  142. model_wrapper.compile(
  143. optimizer=tf.keras.optimizers.Adam(learning_rate=lr_schedule , clipnorm=0.1 ),
  144. loss=keras.losses.MeanSquaredError(),
  146. metrics=['accuracy'])
  148. early_stopping = EarlyStopping(patience=1000, monitor='val_loss' )
  150. mycallbacks = [
  151. early_stopping, #EarlyStopping(patience=10, monitor="loss" ),
  152. tf.keras.callbacks.ModelCheckpoint(
  153. # Path where to save the model
  154. # The two parameters below mean that we will overwrite
  155. # the current checkpoint if and only if
  156. # the `val_loss` score has improved.
  157. # The saved model name will include the current epoch.
  158. filepath="model_{epoch:04d}.keras",
  159. save_best_only= False, #True, # Only save a model if `val_loss` has improved.
  160. monitor="val_loss",
  161. verbose=1
  162. )
  163. ]
  167. history=model_wrapper.fit(train_dataset, epochs=1000, validation_data=(x_val,y_val), callbacks=mycallbacks)
  168. model_wrapper.save("./last_model.keras")
  170. filename = f"trainHistoryDict"
  172. with open(filename, 'wb') as file_pi:
  173. pickle.dump(history.history, file_pi)
  177. model_list = glob.glob('model*.keras')
  178. model_list.sort()
  179. modellength = model_list.__len__()
  180. loss_test = np.zeros(modellength)
  182. for id, mymodel in enumerate(model_list):
  183. model_wrapper = keras.models.load_model(mymodel)
  185. print("Evaluate on test data using ", mymodel )
  186. results = model_wrapper.evaluate(x_test, y_test, batch_size=n1)
  187. print("test loss, test acc:", results)
  188. loss_test[id] = results[0]
  191. id = np.where(loss_test == loss_test.min())[0][0]
  192. model_wrapper = keras.models.load_model(model_list[id])
  193. model_wrapper.save("./best_model.keras")
  195. y2 = model_wrapper.predict(x_test)
  196. y2d = y2 - y_test
  197. sy2d = np.sum(np.absolute(y2d),0)
  198. #mdy = np.mean(np.absolute(y2d),1)
  200. dy = np.absolute(np.array( y2d ) )
  201. two_std = np.std(dy,0) * 2
  202. mdy = np.mean(dy,1)
  203. my2 = np.mean(y2,1)
  204. sdy = np.sum(dy,0)
  205. n = y2.shape[0]
  207. import plotly.express as px
  208. fig = px.box(dy)
  209. #fig.show()
  210. fig.write_html('error.htm')
  211. np.savetxt('parameters_abs_errors.txt', dy )
  214. np.savetxt('test_errors.txt', mdy )
  215. counts, bins = np.histogram(mdy, bins = 100 )
  216. np.savetxt('test_errors_counts.txt', counts)
  217. np.savetxt('test_errors_bins.txt', bins )
  220. fig0, ax0 = plt.subplots()
  221. counts, bins = np.histogram(mdy, bins = 100)
  222. ax0.stairs(counts, bins)
  223. ax0.set_title('Distribution of fitting')
  224. ax0.set_xlabel('average absolute error')
  225. ax0.set_ylabel('amount of samples')
  226. fig0.savefig('ErrorDistribution')
  227. np.savetxt( 'hist_test_error.txt', counts )
  229. fig1, ax1 = plt.subplots()
  230. ax1.stairs(sy2d/n1/2)
  231. ax1.set_title('average absolute test error in each parameter ')
  232. ax1.set_xlabel('parameter #')
  233. ax1.set_ylabel('average absolute error')
  234. fig1.savefig('Errors.png')
  235. np.savetxt('Parameters_mean_errors.txt', sy2d/n1/2 )
  237. x3 = x1 - x1
  238. y3 = model_wrapper.predict(x3)
  239. print("\nPredict Original")
  240. print(y3)
  241. np.savetxt('Predict_Original.txt', y3 )
  243. xn = np.arange(loss_test.size)+1
  244. fig2, ax2 = plt.subplots()
  245. ax2.plot(xn[2:], loss_test[2:])
  246. ax2.plot(xn[2:], history.history['loss'][2:])
  247. ax2.set_title('model loss')
  248. ax2.set_ylabel('loss')
  249. ax2.set_xlabel('epoch')
  250. ax2.legend(['test_loss', 'loss' ], loc='upper left')
  252. fig2.savefig('performance.png')
  253. ett = np.array([xn, loss_test, history.history['loss']]).transpose()
  254. np.savetxt( 'error_test_train.txt', ett )

DL1.py at commit ca8d421, under CC-BY-NC-4.0 · at the source

Overview

Authors: Pei-Chi Yang1,2, Mao-Tsuen Jeng2, Deborah K Lieu3,4, Regan L Smithers3,4, Gonzalo Hernandez-Hernandez1,2, L Fernando Santana2, Colleen E Clancy1,2,5
ORCID iDs: Pei-Chi Yang, L Fernando Santana, Colleen E Clancy
  1. Center for Precision Medicine and Data Science, University of California, Davis, Davis, United States
  2. Department of Physiology and Membrane Biology, University of California, Davis, Davis, United States
  3. Department of Internal Medicine, Division of Cardiovascular Medicine, University of California, Davis, Davis, United States
  4. Institute for Regenerative Cures, University of California, Davis, Davis, United States
  5. Department of Pharmacology, University of California, Davis, Davis, United States
Institutions: University of California, Davis (United States)
Journal: eLife, volume 15, article RP110013
Dates: published online 23 July 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.7554/elife.110013 · PMID 42489674 · PMCID PMC13395456 · OpenAlex W7140182255
Open access: gold, a free copy (OpenAlex)
Status: code verified
Categories: intracellular / patch clamp (modality), human (organism), cellular / molecular (subfield)
Methods: Single-unit activity, calcium imaging
Keywords: None
MeSH: Induced Pluripotent Stem Cells*, Myocytes, Cardiac*, Computer Simulation, Electrophysiological Phenomena, Humans, Neural Networks, Computer, Patch-Clamp Techniques (* major topic)
Topic: Planarian Biology and Electrostimulation (Molecular Biology, Biochemistry, Genetics and Molecular Biology), according to OpenAlex
Funding: University of California, Davis (Chancellor's Postdoctoral Fellowship, Center for Precision Medicine and Data Sciences); NHLBI NIH HHS (R01 HL174001, R01 HL159492, R01 HL128537); NIH HHS (OT2 OD026580); National Heart, Lung, and Blood Institute (R01HL159492, OT2OD026580, R01HL128537, R01HL17400)
Citations: not cited yet (Europe PMC); 60 references in the paper

Abstract

Individual variability shapes how diseases manifest, how patients respond to therapy and how rare phenotypes arise. Conventional experimental approaches obscure variation by averaging which limits mechanistic insight and predictive accuracy. We present a computational framework that builds digital twins of human-induced pluripotent stem cell-derived cardiomyocytes from a single optimized voltage clamp experiment. The framework depends on massive synthetic datasets comprising simulated cells that span broad ionic and electrophysiological ranges. These synthetic data make it possible to control parameters precisely, explore biological variability comprehensively, and train models beyond the limits of experimental data. A neural network trained on synthetic data then inferred biophysical parameters from experimental recordings from live cells, reproducing distinct electrophysiological features. Our study unites computational modeling, data simulation, and learning to enable scalable, precise, individualized cardiac electrophysiology modeling and can be readily extended to any electrically active cell type.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above, with 6 matches between paragraphs and lines of code.

clancylabucd/digital-twin-for-the-win-personalized-cardiac-electrophysiology

License: CC-BY-NC-4.0
State: the link answers, verified on 27 September 2026
Evidence: files inventoried
Commit: ca8d42172eca1dcbf2e98d0ff7b18ff6ab3e2066, 22 December 2025
Languages: C/C++ (6), Python (4), C++ (3), MATLAB (1), Shell (1)
Size: 43 files, 15 scripts
Software Heritage: not archived
Found in: the references
Holds: README, license file
Not found: CITATION.cff, environment file, tests, continuous integration, documentation
Tools: Matplotlib (4 files), NumPy (4 files), Keras (2 files), TensorFlow (2 files), Plotly (1 file)
Availability: 1 check, the latest on 27 September 2026: the link answers
  • 27 September 2026: the link answers
17 files
At the source: archive.softwareheritage.org/swh:1:dir:a279f9489d4603ebf07edfc67faa9169d99e453c;origin=https://github.com/ClancyLabUCD/Digital-Twin-for-the-Win-Personalized-Cardiac-Electrophysiology;visit=swh:1:snp:5354d85e5291451def947dd8e4f774c5fda666a1;anchor=swh:1:rev:ca8d42172eca1dcbf2e98d0ff7b18ff6ab3e2066

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 15 scripts, each with its path and the digest of its content;
  • 6 matches between paragraphs of the paper and lines of the code (method lexical-v1);
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Data availability

Due to the large size of the synthetic datasets (~1.1 M cells), the raw data are not hosted online. However, a representative dataset of 100 cells is provided as an example, and the complete codebase for dataset generation, along with parameter distributions and simulation protocols, is publicly available at GitHub (https://github.com/ClancyLabUCD/Digital-Twin-for-the-Win-Personalized-Cardiac-Electrophysiology; copy archived at Jeng, 2026). This allows full regeneration of the training and validation datasets as described in the study.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, pages, dates, 7 authors, 1 keyword, 7 MeSH terms, 4 funders, 56 references.

Cite

This paper

Yang, P.-C., Jeng, M.-T., Lieu, D. K., Smithers, R. L., Hernandez-Hernandez, G., Santana, L. F., & Clancy, C. E. (2026). Large-scale synthetic data enable digital twins of human excitable cells. eLife, 15, RP110013. https://doi.org/10.7554/elife.110013

BibTeX

@article{yang2026large,
author = {Yang, Pei-Chi and Jeng, Mao-Tsuen and Lieu, Deborah K and Smithers, Regan L and Hernandez-Hernandez, Gonzalo and Santana, L Fernando and Clancy, Colleen E},
title = {{Large-scale synthetic data enable digital twins of human excitable cells}},
journal = {eLife},
year = {2026},
month = jul,
volume = {15},
pages = {RP110013},
publisher = {eLife Sciences Publications, Ltd},
issn = {2050-084X},
doi = {10.7554/elife.110013},
url = {https://doi.org/10.7554/elife.110013},
pmid = {42489674},
pmcid = {PMC13395456}
}

RIS

TY - JOUR
AU - Yang, Pei-Chi
AU - Jeng, Mao-Tsuen
AU - Lieu, Deborah K
AU - Smithers, Regan L
AU - Hernandez-Hernandez, Gonzalo
AU - Santana, L Fernando
AU - Clancy, Colleen E
TI - Large-scale synthetic data enable digital twins of human excitable cells
T2 - eLife
J2 - Elife
PY - 2026
DA - 2026/07/23
VL - 15
SP - RP110013
SN - 2050-084X
PB - eLife Sciences Publications, Ltd
DO - 10.7554/elife.110013
UR - https://doi.org/10.7554/elife.110013
LA - en
ER -

CSL-JSON

{
"id": "10.7554/elife.110013",
"type": "article-journal",
"title": "Large-scale synthetic data enable digital twins of human excitable cells",
"container-title": "eLife",
"author": [
{
"family": "Yang",
"given": "Pei-Chi"
},
{
"family": "Jeng",
"given": "Mao-Tsuen"
},
{
"family": "Lieu",
"given": "Deborah K"
},
{
"family": "Smithers",
"given": "Regan L"
},
{
"family": "Hernandez-Hernandez",
"given": "Gonzalo"
},
{
"family": "Santana",
"given": "L Fernando"
},
{
"family": "Clancy",
"given": "Colleen E"
}
],
"container-title-short": "Elife",
"volume": "15",
"page": "RP110013",
"DOI": "10.7554/elife.110013",
"PMID": "42489674",
"PMCID": "PMC13395456",
"ISSN": "2050-084X",
"publisher": "eLife Sciences Publications, Ltd",
"URL": "https://doi.org/10.7554/elife.110013",
"language": "en",
"issued": {
"date-parts": [
[
2026,
7,
23
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1016/j.premed.2026.100050 [code]
From lipid dynamics to precision predictions: A new approach methodology for precision modeling of phosphoinositide signaling.
Journal: Journal of precision medicine (Amsterdam, Netherlands)
In common: Matplotlib, NumPy, 2 references, author Colleen E Clancy
[2] doi:10.1038/s42003-026-10957-8 [code]
Brain defence by the extracellular matrix protein Cochlin.
Journal: Communications biology
In common: Keras, TensorFlow, Plotly, 2 other tools, cellular / molecular
[3] doi:10.1038/s41467-026-71790-5 [code]
Recurrent DNA break clusters drive replication-stress-induced copy number variants and genome diversification.
Journal: Nature communications
In common: Keras, TensorFlow, Plotly, 2 other tools, cellular / molecular
[4] doi:10.1038/s41467-026-75455-1 [code]
Shared latent representations of speech production for cross-patient speech decoding.
Journal: Nature communications
In common: Keras, TensorFlow, Plotly, 2 other tools
[5] doi:10.1186/s12880-026-02481-2 [code]
Deep learning-based neuroanatomical profiling reveals population-specific brain changes in multiple sclerosis: a large-scale Middle Eastern study.
Journal: BMC medical imaging
In common: Keras, TensorFlow, Plotly, 2 other tools
[6] doi:10.1016/j.xops.2026.101249 [code]
A Disorder-Aware Computational Framework to Identify Structurally Tractable Targets in Proliferative Vitreoretinopathy.
Journal: Ophthalmology science
In common: Keras, TensorFlow, Plotly, 2 other tools
[7] doi:10.1038/s41597-025-05174-7 [code]
A large-scale MEG and EEG dataset for object recognition in naturalistic scenes
Journal: —
In common: Keras, TensorFlow, Plotly, 2 other tools
[8] doi:10.1371/journal.pone.0356243 [code]
Functional organization and natural scene responses across mouse visual cortical areas revealed with encoding manifolds.
Journal: PloS one
In common: Keras, TensorFlow, Plotly, 2 other tools
[9] doi:10.1523/eneuro.0023-26.2026 [code]
Real-Time Segmentation and Classification of Birdsong Syllables for Learning Experiments.
Journal: eNeuro
In common: Keras, TensorFlow, Plotly, 2 other tools
[10] doi:10.1016/j.bpj.2026.06.005 [code]
Curvature-based machine-learning method for automated segmentation of dendritic spines.
Journal: Biophysical journal
In common: Keras, TensorFlow, Plotly, 2 other tools

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.