OSCR

Identification of drug candidates against glioblastoma with machine learning and high-throughput screening of heterogeneous cellular models.

Code ↔ Paper

2 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 2 matches
  1. [1] § Methods › Physicochemical visualisations, model training and computational screen ↔ GBM_ML_virtual_screening.py, lines 86–160 · score 0.98 · Precision Recall curves, colsample_bytree, learning_rate, max_depth, n_estimators, F1 score
  2. [2] § Results › Machine learning models of GBM cell viability ↔ GBM_ML_virtual_screening.py, lines 86–160 · score 0.79 · Monte Carlo loop, virtual screening, XGBoost, accuracy, metrics, precision

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 · 172 lines · 7.6 KB · CC-BY-4.0 · 2 matches

  1. # -*- coding: utf-8 -*-
  2. """
  3. Identification of drug candidates against glioblastoma with machine learning
  4. and high-throughput screening of heterogeneous cellular models:
  6. Machine Learning binary classification code
  7. """
  9. # Libraries
  10. import sys
  11. import pandas as pd
  12. from sklearn.utils import shuffle
  13. from sklearn.ensemble import RandomForestClassifier
  14. from sklearn.pipeline import make_pipeline
  15. from sklearn.preprocessing import MinMaxScaler
  16. import matplotlib.pyplot as plt
  17. import numpy as np
  18. from sklearn.model_selection import train_test_split
  19. import xgboost as xgb
  20. from sklearn.metrics import confusion_matrix
  21. from sklearn.metrics import roc_auc_score
  22. from sklearn.metrics import precision_recall_curve
  23. from sklearn.metrics import auc
  24. import pickle
  26. """
  27. Data reading and preparation. Modify line 33 to exclude from the training procedure any columns that are not the name column,
  28. nor the target column, or the features. If your target column has a different name than apoptosis, change the name of the
  29. feature in lines 39 and 113.
  30. """
  31. def prepare_training_data(df_train):
  33. df_train = df_train.drop(['Salt_eliminated', 'Library', 'SMILES'], axis=1)
  34. df_train = df_train.set_index('Name')
  35. df_train = shuffle(df_train)
  37. # Data split between feature matrix and target array
  38. X_var = df_train.drop(df_train.columns[[0]], axis=1)
  39. y_target = df_train["apoptosis"]
  41. return df_train, X_var, y_target
  43. # Calculation of feature importance using random forest
  44. def feature_importance(X_var_feat_imp, y_feat_imp, df_feat_imp):
  45. rf = RandomForestClassifier()
  46. current = make_pipeline(MinMaxScaler(), rf)
  47. current.fit(X_var_feat_imp,y_feat_imp)
  49. # Feature importance calculation and visualisation
  50. importances_rf = pd.Series(rf.feature_importances_, index = X_var_feat_imp.columns)
  51. sorted_importances_rf = importances_rf.sort_values()
  52. sorted_importances_rf.plot(kind = 'barh', color = 'lightgreen')
  53. plt.xlabel("Feature importance")
  54. plt.ylabel(" ")
  55. plt.yticks(fontsize = 2)
  56. plt.savefig('feature_importance.pdf')
  58. # Write file with feature importance ordered list
  59. sorted_importances_rf.to_csv('rdkit_importance_features.csv')
  61. # Choice of features with importance > 0 for model training
  62. sorted_importances_rf = pd.DataFrame({"Importance" : rf.feature_importances_,
  63. "Name" : X_var_feat_imp.columns})
  64. df_feature_importance = sorted_importances_rf[sorted_importances_rf["Importance"] > 0.0]
  65. important_features = df_feature_importance['Name'].to_list()
  66. print("There are", len(important_features), "features with importance > 0")
  68. #Crop data to include only features with importance > 0
  69. X_var_feat_imp = df_feat_imp[important_features]
  71. return X_var_feat_imp, important_features
  73. # Data preparation for prediction task
  74. def prepare_prediction_data(df_predict, array_importance_feat):
  76. unlabeled_compounds = df_predict
  77. unlabeled_compounds_prob = pd.DataFrame({"Name" : unlabeled_compounds['Name'],
  78. "dummy_column" : np.zeros((len(unlabeled_compounds)),
  79. dtype=int)})
  80. unlabeled_compounds_prob = unlabeled_compounds_prob.set_index('Name')
  81. unlabeled_compounds = unlabeled_compounds.set_index('Name')
  82. X_prep = unlabeled_compounds[array_importance_feat]
  84. return unlabeled_compounds, unlabeled_compounds_prob, X_prep
  86. # Training and prediction Monte Carlo pipeline. The user specifies the number of loops that the code should perform.
  87. def MonteCarlo_loop(user_loop_choice, df_t, X_variant_MonteCarlo, y_train, MonteCarlo_results_prob, X_predict):
  89. Precision_df = []
  90. Recall_df = []
  91. Accuracy_df = []
  92. F1_score_df = []
  93. FPR_df = []
  94. ROC_AUC_df = []
  95. Precision_Recall_AUC_df = []
  96. jj_models_df = []
  98. jj_models = 1
  99. jj = 0
  101. while jj < user_loop_choice:
  103. # Models' training. To add a different model and/or a different parameterization, modify variable model_MonteCarlo in line 105.
  104. X_train_loop, X_test_loop, y_train_loop, y_test_loop = train_test_split(X_variant_MonteCarlo, y_train, test_size = 0.30, stratify = y)
  105. model_MonteCarlo = xgb.XGBClassifier(objective = 'binary:logistic', learning_rate = 0.5, max_depth = 2, n_estimators = 100, colsample_bytree = 0.5)
  106. current_MonteCarlo = make_pipeline(MinMaxScaler(), model_MonteCarlo)
  107. print('Confusion matrix and validation metrics in test set')
  108. current_MonteCarlo.fit(X_train_loop, y_train_loop)
  109. y_pred = current_MonteCarlo.predict(X_test_loop)
  110. accuracy_loop = current_MonteCarlo.score(X_test_loop, y_test_loop)
  111. print(accuracy_loop)
  113. if accuracy_loop > (1 - df_t[df_t["apoptosis"] == 1].shape[0]/df_t.shape[0]):
  114. jj = jj + 1
  116. # Metrics for model
  117. jj_models_df.append(jj_models)
  118. confusion = confusion_matrix(y_test_loop, y_pred)
  119. print(confusion)
  120. accuracy = current_MonteCarlo.score(X_test_loop, y_test_loop)
  121. Accuracy_df.append(accuracy)
  122. FPR = confusion[0,1]/(confusion[0,0] + confusion[0,1])
  123. FPR_df.append(FPR)
  124. Precision = confusion[1,1]/(confusion[1,1] + confusion[0,1])
  125. Precision_df.append(Precision)
  126. Recall = confusion[1,1]/(confusion[1,1] + confusion[1,0])
  127. Recall_df.append(Recall)
  128. F1_score = confusion[1,1]/(confusion[1,1] + 0.5*(confusion[0,1]+confusion[1,0]))
  129. F1_score_df.append(F1_score)
  130. ROC_AUC = roc_auc_score(y_test_loop, y_pred)
  131. ROC_AUC_df.append(ROC_AUC)
  133. # Precision-recall curve
  134. y_test_proba = current_MonteCarlo.predict_proba(X_test_loop)[:,1]
  135. precision, recall, _ = precision_recall_curve(y_test_loop, y_test_proba)
  136. plt.plot(recall, precision, marker='D', color = 'purple', label='Current model')
  137. plt.xlabel('Recall')
  138. plt.ylabel('Precision')
  139. plt.legend()
  140. Precision_Recall_AUC = auc(recall, precision)
  141. Precision_Recall_AUC_df.append(Precision_Recall_AUC)
  143. # Virtual screen
  144. y_proba = current_MonteCarlo.predict_proba(X_predict)[:,1]
  145. MonteCarlo_results_prob[jj_models] = y_proba
  146. filename = ["rsc_gbm_model_", str(jj_models), ".sav"]
  147. filename_arr = "".join(filename)
  148. pickle.dump(current_MonteCarlo, open(str(filename_arr), 'wb'))
  149. jj_models = jj_models + 1
  151. # Screen results
  152. MonteCarlo_results_prob = MonteCarlo_results_prob.drop(MonteCarlo_results_prob.columns[[0]], axis=1)
  153. metrics_mod = pd.DataFrame({"Model" : jj_models_df, "Precision" : Precision_df, "Recall": Recall_df, "Accuracy": Accuracy_df, "F1_score": F1_score_df, "FPR": FPR_df, "ROC_AUC": ROC_AUC_df, "Precision_Recall_AUC": Precision_Recall_AUC_df})
  154. metrics_mod = metrics_mod.set_index('Model')
  156. # Saving parameter and prediction results
  157. metrics_mod.to_csv('rsc_gbm_metrics.csv', sep=",")
  158. MonteCarlo_results_prob.to_csv('rsc_gbm_predictions.csv', sep=",")
  160. return
  162. # Data for ML training
  163. df, X_variant, y = prepare_training_data(pd.read_csv(sys.argv[1]))
  165. # Feature importance analysis
  166. X_variant_after_feat, array_importance_features = feature_importance(X_variant, y, df)
  168. # Data for ML prediction
  169. results, results_prob, X_for_MonteCarlo = prepare_prediction_data(pd.read_csv(sys.argv[2]), array_importance_features)
  171. # Monte Carlo algorithm
  172. MonteCarlo_loop(int(sys.argv[3]), df, X_variant_after_feat, y, results_prob, X_for_MonteCarlo)

GBM_ML_virtual_screening.py, under CC-BY-4.0 · at the source

Overview

Authors: Vanessa Smer-Barreto1,2, Richard J. R. Elliott1, John C. Dawson1, Álvaro Lorente-Macías1, Muhammad Furqan1, Asier Unciti-Broceta1, Diego A. Oyarzún2,3, Neil O. Carragher1
ORCID iDs: Vanessa Smer-Barreto, Richard J. R. Elliott, John C. Dawson, Álvaro Lorente-Macías, Asier Unciti-Broceta, Diego A. Oyarzún
  1. Cancer Research UK Scotland Centre, Institute of Genetics and Cancer, University of Edinburgh Crewe Road South Edinburgh EH4 2XR UK
  2. School of Informatics, University of Edinburgh 10 Crichton St Edinburgh EH8 9AB UK
  3. School of Biological Sciences, University of Edinburgh Max Born Crescent Edinburgh EH9 3BF UK
Institutions: Cancer Research UK (United Kingdom); Edinburgh Cancer Research (United Kingdom); Institute of Genetics and Cancer (United Kingdom); University of Edinburgh (United Kingdom)
Journal: Digital discovery, volume 5, issue 6, pages 2560-2574
Dates: received 9 May 2025; accepted 7 May 2026; published online 13 May 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.1039/d5dd00190k · PMID 42158926 · PMCID PMC13181845 · OpenAlex W4408216194
Open access: diamond, a free copy (OpenAlex)
Preprint: doi:10.1101/2025.03.06.641926
Status: code verified
Categories: other condition (population)
Methods: Connectivity, Preprocessing, Machine learning
Journal subjects: Chemistry
Topic: Computational Drug Discovery Methods (Computational Theory and Mathematics, Computer Science), according to OpenAlex
Funding: Rosetrees Trust (ID2022/100030); Cancer Research UK (CRUK) (C42454/A28596); Brain Tumour Charity (GN-000676); Brain Tumour Research (Unassigned)
Citations: not cited yet (Europe PMC); 69 references in the paper

Abstract

Glioblastoma multiforme (GBM) is an aggressive primary brain tumour that presents significant treatment challenges due to its complex pathology and heterogeneity. The lack of validated molecular targets is a major obstacle for discovering new therapeutic candidates, with no new effective GBM therapies delivered to patients in over two decades. Here, we report the identification of compounds that target the GBM stem cell survival phenotype. Our approach employs machine learning (ML) predictors of cell survival trained on high-throughput, image-based, phenotypic screening data for 3561 compounds, at multiple concentrations, across a panel of six heterogeneous, patient-derived, GBM stem cell lines. We computationally screened more than 12 000 compounds spanning various chemical classes. Experimental validation of ML-identified candidates across the GBM stem cell lines led to the identification of three compounds with activity against the GBM phenotype. Notably, one of our validated hits, the HSP90 inhibitor XL-888, displayed targeted elimination of all six GBM stem cell lines with IC50 in the nanomolar range. Further analyses suggest an XL-888 mechanism of action based on competitive ATP inhibition of HSP90 followed by disruption of HSP90 client proteins, and identify XL-888 as a promising candidate for future personalised medicine campaigns. Our work demonstrates that the use of phenotypic screening in tandem with ML can effectively identify therapeutic leads for personalised treatments in highly heterogeneous indications with few known molecular targets.

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 2 matches between paragraphs and lines of code.

Zenodo 17100377

License: CC-BY-4.0
State: the link answers, verified on 28 September 2026
Evidence: files inventoried
Languages: Python (1)
Size: 3 files, 1 script
Software Heritage: not checked
Found in: “Data availability”
Not found: README, license file, CITATION.cff, environment file, tests, continuous integration, documentation
Tools: Matplotlib (1 file), NumPy (1 file), pandas (1 file), scikit-learn (1 file), XGBoost (1 file)
Availability: 1 check, the latest on 28 September 2026: the link answers (HTTP 200)
  • 28 September 2026: the link answers (HTTP 200)
1 file
At the source: doi.org/10.5281/zenodo.17100377

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;
  • 1 script, each with its path and the digest of its content;
  • 2 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

Training and screening data have been deposited in Zenodo at https://doi.org/10.5281/zenodo.17100377.

Code availability: python code for model training and compound screening has been deposited in Zenodo at https://doi.org/10.5281/zenodo.17100377.

Supplementary information (SI): representative images of the six glioma stem cell lines used in this work; the experimetal results concern HSP90 inhibitors' studies of apoptosis induction, cell cycle and stemness effects; a structural and physicochemical properties analysis of the compound XL-888. See DOI: https://doi.org/10.1039/d5dd00190k.

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 2, 28 September 2026

  • Publisher: n/a → Royal Society of Chemistry

Version 1, 28 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 8 authors, 4 funders, 63 references.

Cite

This paper

Smer-Barreto, V., Elliott, R. J. R., Dawson, J. C., Lorente-Macías, Á., Furqan, M., Unciti-Broceta, A., Oyarzún, D. A., & Carragher, N. O. (2026). Identification of drug candidates against glioblastoma with machine learning and high-throughput screening of heterogeneous cellular models. Digital discovery, 5(6), 2560-2574. https://doi.org/10.1039/d5dd00190k

BibTeX

@article{smerbarreto2026identification,
author = {Smer-Barreto, Vanessa and Elliott, Richard J. R. and Dawson, John C. and Lorente-Macías, Álvaro and Furqan, Muhammad and Unciti-Broceta, Asier and Oyarzún, Diego A. and Carragher, Neil O.},
title = {{Identification of drug candidates against glioblastoma with machine learning and high-throughput screening of heterogeneous cellular models}},
journal = {Digital discovery},
year = {2026},
month = may,
volume = {5},
number = {6},
pages = {2560--2574},
publisher = {Royal Society of Chemistry},
issn = {2635-098X},
doi = {10.1039/d5dd00190k},
url = {https://doi.org/10.1039/d5dd00190k},
pmid = {42158926},
pmcid = {PMC13181845}
}

RIS

TY - JOUR
AU - Smer-Barreto, Vanessa
AU - Elliott, Richard J. R.
AU - Dawson, John C.
AU - Lorente-Macías, Álvaro
AU - Furqan, Muhammad
AU - Unciti-Broceta, Asier
AU - Oyarzún, Diego A.
AU - Carragher, Neil O.
TI - Identification of drug candidates against glioblastoma with machine learning and high-throughput screening of heterogeneous cellular models
T2 - Digital discovery
J2 - Digit Discov
PY - 2026
DA - 2026/05/13
VL - 5
IS - 6
SP - 2560
EP - 2574
SN - 2635-098X
PB - Royal Society of Chemistry
DO - 10.1039/d5dd00190k
UR - https://doi.org/10.1039/d5dd00190k
LA - en
ER -

CSL-JSON

{
"id": "10.1039/d5dd00190k",
"type": "article-journal",
"title": "Identification of drug candidates against glioblastoma with machine learning and high-throughput screening of heterogeneous cellular models",
"container-title": "Digital discovery",
"author": [
{
"family": "Smer-Barreto",
"given": "Vanessa"
},
{
"family": "Elliott",
"given": "Richard J. R."
},
{
"family": "Dawson",
"given": "John C."
},
{
"family": "Lorente-Macías",
"given": "Álvaro"
},
{
"family": "Furqan",
"given": "Muhammad"
},
{
"family": "Unciti-Broceta",
"given": "Asier"
},
{
"family": "Oyarzún",
"given": "Diego A."
},
{
"family": "Carragher",
"given": "Neil O."
}
],
"container-title-short": "Digit Discov",
"volume": "5",
"issue": "6",
"page": "2560-2574",
"DOI": "10.1039/d5dd00190k",
"PMID": "42158926",
"PMCID": "PMC13181845",
"ISSN": "2635-098X",
"publisher": "Royal Society of Chemistry",
"URL": "https://doi.org/10.1039/d5dd00190k",
"language": "en",
"issued": {
"date-parts": [
[
2026,
5,
13
]
]
}
}

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:10.3390/medicina62061169 [code]
Survival Outcomes and Machine Learning-Based Prediction of 12-Month Mortality in Glioblastoma Before and During the COVID-19 Pandemic: A SEER Population-Based Study.
Journal: Medicina (Kaunas, Lithuania)
In common: XGBoost, scikit-learn, pandas, 2 other tools, other condition, 1 reference
[2] doi:10.1126/sciadv.aeg3223 [code]
The extreme diversity of retinal amacrine cells has deep evolutionary roots.
Journal: Science advances
In common: XGBoost, scikit-learn, pandas, 2 other tools, 1 reference
[3] doi:10.1016/j.xcrm.2026.102766 [code]
A longitudinal single-cell and spatial multiomic atlas of pediatric high-grade glioma.
Journal: Cell reports. Medicine
In common: scikit-learn, pandas, Matplotlib, 1 other tool, other condition, 2 references
[4] doi:10.3389/fonc.2026.1750334 [code]
Machine-learning assisted subclassification of glioblastoma by developing an endoplasmic reticulum stress-related methylation signature.
Journal: Frontiers in oncology
In common: XGBoost, scikit-learn, pandas, 1 other tool, other condition, 1 reference
[5] doi:10.1093/neuonc/noag068 [code]
PTBP1 knockdown reprograms glioma stem cells into neuronal-like cells and suppresses tumorigenesis via the DUSP5-ERK1/2 signaling pathway.
Journal: Neuro-oncology
In common: scikit-learn, pandas, Matplotlib, 1 other tool, other condition, 2 references
[6] doi:10.1016/j.cell.2026.05.026 [code]
The critical role of the endogenous immune compartment after CAR T cell therapy in recurrent GBM.
Journal: Cell
In common: scikit-learn, pandas, Matplotlib, 1 other tool, other condition, 2 references
[7] doi:10.3389/fncel.2026.1713437 [code]
Single-nucleus transcriptomics identifies cell cycle and synaptic pathway dysregulation during OPC-to-glioma progression.
Journal: Frontiers in cellular neuroscience
In common: other condition, 4 references
[8] doi:10.1186/s13244-026-02365-7 [code]
Super-resolution MRI and 2.5D deep learning for intratumoral-peritumoral radiomics in preoperative prediction of rectal cancer perineural invasion.
Journal: Insights into imaging
In common: XGBoost, scikit-learn, pandas, 2 other tools, other condition
[9] doi:10.1038/s41467-026-76837-1 [code]
Drug screen and machine learning predict neuroprotective agents in a preclinical human model of childhood dementia.
Journal: Nature communications
In common: XGBoost, scikit-learn, pandas, 2 other tools, other condition
[10] doi:10.1038/s41598-026-64405-y [code]
Evaluating clinical and neuroimaging predictors for cognitive-behavioral therapy outcome in obsessive-compulsive disorder.
Journal: Scientific reports
In common: XGBoost, scikit-learn, pandas, 2 other tools, other condition

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.