Machine Learning Approaches to Classify Primary and Metastatic Cancers Using Tissue of Origin-Based DNA Methylation Profiles.

{"sci_version": "1.2", "schema": "https://corpus.bntrg.co.uk/sci/1.0/sci.schema.json", "id": "pmid:34359669", "kind": "science-record", "revision": 8, "content_hash": "sha256:087ad3dc2be47ed9839db085e9239cdcc0d75c4d693997e8042a697c16f049fa", "created_at": "2026-11-01T15:41:47Z", "updated_at": "2026-11-01T22:47:17Z", "generator": {"name": "corpus-assay-engine", "run_id": "reading-sci-2026-11-01"}, "paper": {"state": "complete", "title": "Machine Learning Approaches to Classify Primary and Metastatic Cancers Using Tissue of Origin-Based DNA Methylation Profiles.", "doi": "10.3390/cancers13153768", "pmid": "34359669", "authors": ["Vijayachitra Modhukur", "Shakshi Sharma", "Mainak Mondal", "Ankita Lawarde", "Keiu Kask", "Rajesh Sharma", "Andres Salumets"], "journal": "Cancers", "year": 2021, "abstract": "Metastatic cancers account for up to 90% of cancer-related deaths. The clear differentiation of metastatic cancers from primary cancers is crucial for cancer type identification and developing targeted treatment for each cancer type. DNA methylation patterns are suggested to be an intriguing target for cancer prediction and are also considered to be an important mediator for the transition to metastatic cancer. In the present study, we used 24 cancer types and 9303 methylome samples downloaded from publicly available data repositories, including The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO). We constructed machine learning classifiers to discriminate metastatic, primary, and non-cancerous methylome samples. We applied support vector machines (SVM), Naive Bayes (NB), extreme gradient boosting (XGBoost), and random forest (RF) machine learning models to classify the cancer types based on their tissue of origin. RF outperformed the other classifiers, with an average accuracy of 99%. Moreover, we applied local interpretable model-agnostic explanations (LIME) to explain important methylation biomarkers to classify cancer types.", "claims": [{"id": "claim-method", "statement": "A linear model fit was performed using the Bioconductor (version number 3.13) package limma (version number 3.48.0) [28] to identify differentially methylated CpG sites (DMS).", "kind": "methodological", "target": null, "evidence": [{"asset": "source:pmid:34359669", "locator": "sections[5] \"2.3. Identification of Methylation-Based Cancer Biomarkers\" paragraph 2", "quote": "A linear model fit was performed using the Bioconductor (version number 3.13) package limma (version number 3.48.0) [28] to identify differentially methylated CpG sites (DMS)."}], "state": "complete"}, {"id": "claim-data-access", "statement": "The data that support the findings of this study are openly available in the TCGA database accessible via the Xena Browser (https://xena.ucsc.edu/, last accessed 1 April 2021). Source code used in this study is publicly available in a Github repository (https://github.com/shakshi12/AI-Explainability---BioInformatics, accessed on 15 May 2021. A Jupyter Notebook to replicate all the machine learning experiments used in the study is also available in the Github repository.", "kind": "methodological", "target": null, "evidence": [{"asset": "source:pmid:34359669", "locator": "sections[28] \"Data Availability Statement\" paragraph 2", "quote": "The data that support the findings of this study are openly available in the TCGA database accessible via the Xena Browser (https://xena.ucsc.edu/, last accessed 1 April 2021). Source code used in this study is publicly available in a Github repository (https://github.com/shakshi12/AI-Explainability---BioInformatics, accessed on 15 May 2021. A Jupyter Notebook to replicate all the machine learning experiments used in the study is also available in the Github repository."}], "state": "complete"}], "gaps": []}, "experiment": {"state": "partial", "objective": null, "steps": [{"id": "read_published_estimate", "order": 1, "name": "Read the published estimate from the source", "kind": "other", "state": "complete", "blocked_on": null, "description": "The claim, method and evidence span published with the source article.", "duration_hours": null, "parameters": {}, "inputs": ["source:pmid:34359669"], "outputs": ["the published estimate and its evidence"], "instruments": [], "reagents": []}, {"id": "obtain_source_data", "order": 2, "name": "Obtain the source data from the named repository", "kind": "data-preparation", "state": "blocked", "blocked_on": "data:accession_absent", "description": "The GEO series accession resolves and its sample accessions are listed. This lane did not download the dataset or read its licence.", "duration_hours": null, "parameters": {}, "inputs": [], "outputs": ["the source data with its repository provenance"], "instruments": [], "reagents": []}, {"id": "run_analysis", "order": 3, "name": "Run the published analysis", "kind": "analysis", "state": "complete", "blocked_on": null, "description": "Apply the paper's statistical method to the resolved series and compare the result with the published target.", "duration_hours": null, "parameters": {}, "inputs": [], "outputs": ["the replicated result and its comparison"], "instruments": [], "reagents": []}], "materials": [], "instruments": [], "reagents": [], "conditions": {}, "gaps": ["step:obtain_source_data blocked on data:accession_absent"]}, "data": {"state": "partial", "assets": [{"id": "source:pmid:34359669", "role": "manuscript", "format": "xml", "media_type": "application/xml", "uri": "https://doi.org/10.3390/cancers13153768", "sha256": "sha256:1754456add2c6f9961edcd6331c9baa21a4478b337da4d4c8442a404607433d0", "bytes": 50999, "access": "open", "state": "resolved", "reader": null, "reader_status": "reader_required", "identifier": "10.3390/cancers13153768", "version": null, "note": "Full article fetched from PMC; the checksum covers the rendered section body, not the original XML bytes.", "structure": {"pmcid": "PMC8345047", "source": "europe_pmc_fulltext", "coverage": "whole_body", "text_hash_scope": "rendered_section_body"}}, {"id": "asset:table:pmid:34359669:1", "role": "table", "format": "jats-table", "media_type": "application/vnd.jats.table+json", "uri": null, "sha256": "sha256:d66a57757669f4fcb2e942ff3221baaca3d340b973de014668b68ffc6c0acd7c", "bytes": 4584, "access": "open", "state": "verified", "reader": "read_table", "reader_status": "available", "identifier": "section[16]/table-wrap[1]#cancers-13-03768-t001", "version": null, "note": "Performance of the RF multi-label classification model after performing SMOTE for the prediction of every tissue/cancer type. Tissue/cancer type prediction, tissue origin, and labels are included in the table, along with the performance metrics. BLCA: bladder urothelial carcinoma; BRCA: breast invasive carcinoma; CESC: cervical squamous cell carcinoma and endocervical adenocarcinoma; CHOL: cholangiocarcinoma; COAD: colon adenocarcinoma; ESCA: esophageal carcinoma; GBM: glioblastoma multiforme; HNSC: head and neck squamous cell carcinoma; KIRC: kidney renal clear cell carcinoma; KIRP: kidney renal papillary cell carcinoma; LIHC: liver hepatocellular carcinoma; LUAD: lung adenocarcinoma; LUSC: lung squamous cell carcinoma; OV: ovarian cancer; PAAD: pancreatic adenocarcinoma; PCPG: pheochromocytoma and paraganglioma; PRAD: prostate adenocarcinoma; READ: rectum adenocarcinoma; SARC: sarcoma; SKCM: skin cutaneous melanoma; STAD: stomach adenocarcinoma; THCA: thyroid carcinoma; THYM: thymoma; UCEC: uterine corpus endometrial carcinoma.", "structure": {"pmcid": "PMC8345047", "section_index": 16, "section_title": "3.3. Classifier Performance after Balancing the Methylome Data Sets Using SMOTE", "row_count": 51, "column_count": 7}, "table": {"label": "Table 1", "caption": "Performance of the RF multi-label classification model after performing SMOTE for the prediction of every tissue/cancer type. Tissue/cancer type prediction, tissue origin, and labels are included in the table, along with the performance metrics. BLCA: bladder urothelial carcinoma; BRCA: breast invasive carcinoma; CESC: cervical squamous cell carcinoma and endocervical adenocarcinoma; CHOL: cholangiocarcinoma; COAD: colon adenocarcinoma; ESCA: esophageal carcinoma; GBM: glioblastoma multiforme; HNSC: head and neck squamous cell carcinoma; KIRC: kidney renal clear cell carcinoma; KIRP: kidney renal papillary cell carcinoma; LIHC: liver hepatocellular carcinoma; LUAD: lung adenocarcinoma; LUSC: lung squamous cell carcinoma; OV: ovarian cancer; PAAD: pancreatic adenocarcinoma; PCPG: pheochromocytoma and paraganglioma; PRAD: prostate adenocarcinoma; READ: rectum adenocarcinoma; SARC: sarcoma; SKCM: skin cutaneous melanoma; STAD: stomach adenocarcinoma; THCA: thyroid carcinoma; THYM: thymoma; UCEC: uterine corpus endometrial carcinoma.", "rows": [["Label", "Tissue Origin", "Cancer Type Prediction (Normal Tissue/Primary Cancer/Metastatic Cancer)", "Precision", "Recall", "F1-Score", "Accuracy"], ["BLCA.Normal", "Bladder", "Normal", "1.0", "1.0", "1.0", "1.0"], ["BLCA.Primary", "Bladder", "Primary", "0.994", "0.956", "0.974", "1.0"], ["BRCA.Metastatic", "Breast", "Metastatic", "0.973", "1.0", "0.986", "1.0"], ["BRCA.Normal", "Breast", "Normal", "0.987", "1.0", "0.993", "1.0"], ["BRCA.Primary", "Breast", "Primary", "0.976", "0.932", "0.954", "0.926"], ["CESC.Primary", "Cervix", "Primary", "0.971", "0.982", "0.977", "0.952"], ["CHOL.Normal", "Bile duct", "Normal", "1.0", "1.0", "1.0", "1.0"], ["CHOL.Primary", "Bile duct", "Primary", "1.0", "1.0", "1.0", "1.0"], ["COAD.Metastatic", "Colon and/or rectum", "Metastatic", "1.0", "1.0", "1.0", "1.0"], ["COAD.Normal", "Colon and/or rectum", "Normal", "1.0", "1.0", "1.0", "1.0"], ["COAD.Primary", "Colon and/or rectum", "Primary", "0.993", "0.980", "0.987", "0.98"], ["ESCA.Normal", "Esophagus", "Normal", "1.0", "1.0", "1.0", "1.0"], ["ESCA.Primary", "Esophagus", "Primary", "0.982", "0.921", "0.951", "0.975"], ["GBM.Normal", "Brain", "Normal", "1.0", "1.0", "1.0", "1.0"], ["GBM.Primary", "Brain", "Primary", "1.0", "1.0", "1.0", "1.0"], ["HNSC.Normal", "Head and Neck", "Normal", "0.994", "1.0", "0.997", "1.0"], ["HNSC.Primary", "Head and Neck", "Primary", "0.921", "0.926", "0.924", "0.950"], ["KIRC.Normal", "Kidney", "Normal", "1.0", "0.993", "0.996", "1.0"], ["KIRC.Primary", "Kidney", "Primary", "1.0", "0.974", "0.987", "0.993"], ["KIRP.Normal", "Kidney", "Normal", "0.993", "1.0", "0.996", "1.0"], ["KIRP.Primary", "Kidney", "Primary", "0.986", "1.0", "0.993", "0.989"], ["LIHC.Normal", "Liver", "Normal", "1.0", "1.0", "1.0", "1.0"], ["LIHC.Primary", "Liver", "Primary", "1.0", "0.993", "0.996", "0.996"], ["LUAD.Normal", "Lung", "Normal", "1.0", "1.0", "1.0", "1.0"], ["LUAD.Primary", "Lung", "Primary", "0.993", "0.993", "0.993", "0.974"], ["LUSC.Normal", "Lung", "Normal", "0.993", "1.0", "0.996", "1.0"], ["LUSC.Primary", "Lung", "Primary", "0.910", "0.953", "0.931", "0.951"], ["OV.Metastatic", "Ovary", "Metastatic", "1.0", "1.0", "1.0", "1.0"], ["OV.Normal", "Ovary", "Normal", "1.0", "1.0", "1.0", "1.0"], ["OV.Primary", "Ovary", "Primary", "1.0", "1.0", "1.0", "1.0"], ["PAAD.Normal", "Pancreas", "Normal", "1.0", "1.0", "1.0", "1.0"], ["PAAD.Primary", "Pancreas", "Primary", "0.993", "0.986", "0.989", "0.981"], ["PCPG.Primary", "Adrenal gland", "Primary", "1.0", "1.0", "1.0", "1.0"], ["PRAD.Metastatic", "Prostate", "Metastatic", "1.0", "1.0", "1.0", "1.0"], ["PRAD.Normal", "Prostate", "Normal", "0.980", "1.0", "0.990", "0.982"], ["PRAD.Primary", "Prostate", "Primary", "1.0", "0.980", "0.989", "0.988"], ["READ.Normal", "Rectum", "Normal", "1.0", "1.0", "1.0", "1.0"], ["READ.Primary", "Rectum", "Primary", "0.992", "1.0", "0.996", "1.0"], ["SARC.Primary", "Soft tissue", "Primary", "0.970", "1.0", "0.985", "1.0"], ["SKCM.Metastatic", "Skin", "Metastatic", "0.986", "0.980", "0.983", "0.974"], ["SKCM.Primary", "Skin", "Primary", "0.980", "1.0", "0.990", "0.993"], ["STAD.Primary", "Stomach", "Primary", "0.944", "0.964", "0.954", "0.939"], ["THCA.Metastatic", "Thyroid", "Metastatic", "1.0", "1.0", "1.0", "1.0"], ["THCA.Normal", "Thyroid", "Normal", "1.0", "1.0", "1.0", "1.0"], ["THCA.Primary", "Thyroid", "Primary", "1.0", "1.0", "1.0", "0.993"], ["THYM.Primary", "Thymus", "Primary", "1.0", "1.0", "1.0", "1.0"], ["UCEC.Metastatic", "Uterus", "Metastatic", "0.971", "1.0", "0.985", "1.0"], ["UCEC.Normal", "Uterus", "Normal", "0.993", "1.0", "0.996", "1.0"], ["UCEC.Primary", "Uterus", "Primary", "0.987", "0.970", "0.978", "0.961"], ["Average accuracy", "", "", "", "", "", "0.981"]], "row_count": 51, "column_count": 7, "numeric_cells": 197, "header_rows": 1, "locator": "section[16]/table-wrap[1]#cancers-13-03768-t001"}}], "gaps": []}, "analysis": {"state": "partial", "methods": [{"id": "method-1", "family": "differential_expression", "specific": "limma_voom", "label": "Differential expression analysis", "dsl_operations": ["normalise_counts", "differential_expression_fit", "fdr_bh"], "missing_operations": [], "expressible": true, "parameters": {}, "data_shapes": ["count_matrix_with_condition"], "evidence": {"score": 1.0, "band": "verbatim", "quote": "A linear model fit was performed using the Bioconductor (version number 3.13) package limma (version number 3.48.0) [28] to identify differentially methylated CpG sites (DMS)."}}, {"id": "method-2", "family": "principal_component_analysis", "specific": "pca", "label": "Principal component analysis", "dsl_operations": ["pca_fit"], "missing_operations": [], "expressible": true, "parameters": {}, "data_shapes": ["wide_numeric_matrix"], "evidence": {"score": 0.8, "band": "high", "quote": "Principal component analysis"}}, {"id": "method-3", "family": "machine_learning", "specific": "random_forest", "label": "Machine learning", "dsl_operations": [], "missing_operations": ["machine_learning_random_forest"], "expressible": false, "parameters": {}, "data_shapes": ["labelled_feature_matrix"], "evidence": {"score": 0.7, "band": "medium", "quote": "random forest"}}, {"id": "method-4", "family": "machine_learning", "specific": "support_vector_machine", "label": "Machine learning", "dsl_operations": ["train_classifier", "cross_validate", "roc_auc"], "missing_operations": [], "expressible": true, "parameters": {}, "data_shapes": ["labelled_feature_matrix"], "evidence": {"score": 0.7, "band": "medium", "quote": "support vector machines"}}, {"id": "method-5", "family": "machine_learning", "specific": "gradient_boosting", "label": "Machine learning", "dsl_operations": ["train_classifier", "cross_validate", "roc_auc"], "missing_operations": [], "expressible": true, "parameters": {}, "data_shapes": ["labelled_feature_matrix"], "evidence": {"score": 0.7, "band": "medium", "quote": "gradient boosting"}}, {"id": "method-6", "family": "network_analysis", "specific": "ppi_network", "label": "Network / pathway analysis", "dsl_operations": ["build_graph", "centrality_scores", "pathway_enrich"], "missing_operations": [], "expressible": true, "parameters": {}, "data_shapes": ["edge_list"], "evidence": {"score": 0.8, "band": "high", "quote": "PPI"}}], "targets": [], "gaps": ["method.reader_differs_from_extractor"]}, "capability": {"state": "complete", "requirements": [{"capability_id": "capability:computational_analysis", "label": "Computational analysis platform", "assay_families": ["meta_analysis", "machine_learning", "survival_analysis", "statistical_regression", "pca", "molecular_docking", "network_analysis", "differential_expression", "gwas"], "availability_state": "owned", "tier_unlocked": "R1", "candidate": "Automated Assays Laravel application and simulated adapters"}], "gaps": []}, "readiness": {"state": "blocked", "tier": "R4", "blocking_axis": "data", "blocked_on": "data:accession_absent", "axes": {"method": {"tier": "R1", "expressible": true, "required_operations": ["normalise_counts", "differential_expression_fit", "fdr_bh"], "missing_operations": [], "basis": "The paper names the method it used and the corpus's method extractor maps it onto the DSL catalogue: limma_voom."}, "capability": {"tier": "R1", "capability_id": "capability:computational_analysis", "availability_state": "owned", "basis": "The computational analysis is owned. The physical bench the paper used is not, so a physical validation cannot be run here."}, "data": {"tier": "R4", "reachable": false, "accessions": [], "resolved_accessions": [], "statements": ["The data that support the findings of this study are openly available in the TCGA database accessible via the Xena Browser (https://xena.ucsc.edu/, last accessed 1 April 2021). Source code used in this study is publicly available in a Github repository (https://github.com/shakshi12/AI-Explainability---BioInformatics, accessed on 15 May 2021. A Jupyter Notebook to replicate all the machine learning experiments used in the study is also available in the Github repository."], "basis": "The paper names no accession to its source data, so the data has to come from the authors."}}, "can_run": {"dsl_operations": ["normalise_counts", "differential_expression_fit", "fdr_bh"], "assets": ["source:pmid:34359669", "asset:table:pmid:34359669:1"], "steps": ["run_analysis"]}, "must_build": {"dsl_operations": [], "converters": [], "readers": [], "capabilities": [], "data_gaps": ["data:geo:None"]}, "gaps": ["data:accession_absent"]}, "review": {"state": "partial", "machine_critique": {"kind": "extraction_review", "reader": {"model": "MiniMax-M3.1-Flash", "provider": "minimax", "provider_family": "minimax", "read_at": "2026-10-31T18:33:39Z", "independence": "single_reader"}, "evidence_check": {"method": "verbatim", "availability": "verbatim"}, "calls": 3, "usage": {"prompt_tokens": 39019, "completion_tokens": 2041, "cached_tokens": 776}}, "human_critique": null, "stages": [{"id": "extraction", "state": "complete", "reader": "MiniMax-M3.1-Flash", "provider": "minimax"}], "gaps": ["review.independent_reader", "review.machine_critique", "review.human_critique"]}, "execution": {"state": "absent", "status": "not_run", "mode": null, "runs": [], "gaps": ["execution.not_run"]}, "provenance": {"entities": [{"id": "pmid:34359669", "type": "paper", "label": "pmid:34359669"}, {"id": "source:pmid:34359669", "type": "manuscript", "label": "PMID 34359669 full text"}, {"id": "data:geo:None", "type": "dataset", "label": "GEO None"}], "activities": [{"id": "activity-reading-sci-2026-11-01", "type": "classification", "used": [{"id": "pmid:34359669", "type": "paper", "label": "pmid:34359669"}, {"id": "source:pmid:34359669", "type": "manuscript", "label": "PMID 34359669 full text"}, {"id": "data:geo:None", "type": "dataset", "label": "GEO None"}], "agent": "agent:corpus-assay-engine"}, {"id": "activity-read-34359669", "type": "extraction", "used": ["source:pmid:34359669"], "agent": "agent:minimax-reader"}, {"id": "activity-resolve-None", "type": "accession_resolution", "used": ["data:geo:None"], "agent": "agent:corpus-assay-engine"}], "agents": [{"id": "agent:corpus-assay-engine", "type": "software", "label": "Corpus assay engine"}, {"id": "agent:minimax-reader", "type": "software", "label": "MiniMax literature reader"}], "source_run_id": "reading-sci-2026-11-01", "dsl_catalogue_fingerprint": "343fb30756ba6ec3", "capability_table_fingerprint": "2cb3e42f047d2ecf", "snapshot_id": "ba34b49490d9bfaf"}, "lineage": {"parent_id": "pmid:34359669", "parent_hash": "sha256:020ef3725a37dfb8f679f1f82b782dc9c68ae283f2da62a0ecdda6ee0d4fe4e0", "fork_of": null, "generation": 7}, "gaps": ["execution.not_run", "method.reader_differs_from_extractor", "review.human_critique", "review.independent_reader", "review.machine_critique", "step:obtain_source_data blocked on data:accession_absent"], "assays": [{"role": "analysis", "definition_id": "differential_expression", "definition_version": "f2132e96ae6c7a4e", "definition_digest": null, "required_operations": ["normalise_counts", "differential_expression_fit", "fdr_bh"], "support_state": "supported", "source_hashes": {"dsl_catalogue": "f2132e96ae6c7a4e"}, "confidence": 1.0, "evidence": ["A linear model fit was performed using the Bioconductor (version number 3.13) package limma (version number 3.48.0) [28] to identify differentially methylated CpG sites (DMS)."], "order_basis": "analysis_only", "id": "assay:1"}]}