Network Organization, Null-Model Assessment, and Robustness of a Published Mycobacterium tuberculosis Candidate Drug-Resistance Gene Panel

Drug-resistant tuberculosis remains a major challenge, but resistance-associated genes are commonly interpreted as separate drug-specific lists. We reorganized a previously published TBProfiler Tier 1/Tier 2 candidate-gene panel into a reproducible bipartite network and quantified connected components, bipartite density, centrality, Jaccard similarity, the Szymkiewicz–Simpson overlap coefficient, evidence-tier sensitivity, degree-preserving null expectations, and leave-one-edge/gene-out robustness. The full network contained 14 drug/drug-class nodes, 49 genes, 71 edges, six disconnected components, and a bipartite density of 0.1035. In the full network, rrs had the highest number (four) of linked drugs, whereas Rv1258c had the highest full-network betweenness (0.203) but fell to zero in the Tier 1-only network. In 10,000 degree-preserving randomizations, however, Rv1258c betweenness did not exceed the null expectation (empirical p = 0.205), indicating that this ranking is compatible with the panel’s degree structure. Pairwise drug similarity, as expressed by the Jaccard coefficient, highlighted bedaquiline-clofazimine (0.833), amikacin-capreomycin (0.625), amikacin-kanamycin (0.429), and isoniazid-ethionamide (0.300); corresponding Szymkiewicz–Simpson overlap coefficients were 1.000, 0.833, 1.000, and 0.500, respectively. In the primary analysis encompassing all 91 drug-pair comparisons, three overlaps—bedaquiline–clofazimine, amikacin–capreomycin, and amikacin–kanamycin—exceeded degree-preserving null expectations after BH-FDR correction (q = 0.0046, 0.0046, and 0.0243, respectively). The separate analysis restricted to the 12 pairs with non-zero observed Jaccard similarity was considered exploratory and did not inform the primary conclusions. The amikacin–capreomycin Jaccard value fell to 0.250 under Tier 1 restriction, showing dependence on Tier 2 assignments. Collapsing amikacin, kanamycin, and capreomycin into a single second-line-injectable node preserved six components and broadly retained gene-centrality ranks (Spearman’s rank correlation coefficient, ρ= 0.851). Published CRyPTIC co-resistance patterns remained concordant with the bedaquiline–clofazimine and amikacin–kanamycin overlaps. The framework therefore quantifies the organization, evidence-tier dependence, and robustness of existing candidate-resistance associations; it does not establish novel resistance mechanisms, diagnostic accuracy, or patient-level clinical predictions.

Authors

Institutions

Publication Details

Journal
International Journal of Molecular Sciences
Published
2026-11-20
DOI
https://doi.org/10.3390/ijms272210454
Primary Topic
Tuberculosis Research and Epidemiology
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Network Organization, Null-Model Assessment, and Robustness of a Published Mycobacterium tuberculosis Candidate Drug-Resistance Gene Panel

Ayşe Emel Önal, Uğurcan Sayılı, Halim İşsever, Gözde Öztan et al.
1 links
International Journal of Molecular Sciences
Tuberculosis Research and Epidemiology
article

Network Organization, Null-Model Assessment, and Robustness of a Published Mycobacterium tuberculosis Candidate Drug-Resistance Gene Panel

Ayşe Emel Önal, Uğurcan Sayılı, Halim İşsever, Gözde Öztan, Mehmet Burhan Küçükoğlu
article en

Abstract

Drug-resistant tuberculosis remains a major challenge, but resistance-associated genes are commonly interpreted as separate drug-specific lists. We reorganized a previously published TBProfiler Tier 1/Tier 2 candidate-gene panel into a reproducible bipartite network and quantified connected components, bipartite density, centrality, Jaccard similarity, the Szymkiewicz–Simpson overlap coefficient, evidence-tier sensitivity, degree-preserving null expectations, and leave-one-edge/gene-out robustness. The full network contained 14 drug/drug-class nodes, 49 genes, 71 edges, six disconnected components, and a bipartite density of 0.1035. In the full network, rrs had the highest number (four) of linked drugs, whereas Rv1258c had the highest full-network betweenness (0.203) but fell to zero in the Tier 1-only network. In 10,000 degree-preserving randomizations, however, Rv1258c betweenness did not exceed the null expectation (empirical p = 0.205), indicating that this ranking is compatible with the panel’s degree structure. Pairwise drug similarity, as expressed by the Jaccard coefficient, highlighted bedaquiline-clofazimine (0.833), amikacin-capreomycin (0.625), amikacin-kanamycin (0.429), and isoniazid-ethionamide (0.300); corresponding Szymkiewicz–Simpson overlap coefficients were 1.000, 0.833, 1.000, and 0.500, respectively. In the primary analysis encompassing all 91 drug-pair comparisons, three overlaps—bedaquiline–clofazimine, amikacin–capreomycin, and amikacin–kanamycin—exceeded degree-preserving null expectations after BH-FDR correction (q = 0.0046, 0.0046, and 0.0243, respectively). The separate analysis restricted to the 12 pairs with non-zero observed Jaccard similarity was considered exploratory and did not inform the primary conclusions. The amikacin–capreomycin Jaccard value fell to 0.250 under Tier 1 restriction, showing dependence on Tier 2 assignments. Collapsing amikacin, kanamycin, and capreomycin into a single second-line-injectable node preserved six components and broadly retained gene-centrality ranks (Spearman’s rank correlation coefficient, ρ= 0.851). Published CRyPTIC co-resistance patterns remained concordant with the bedaquiline–clofazimine and amikacin–kanamycin overlaps. The framework therefore quantifies the organization, evidence-tier dependence, and robustness of existing candidate-resistance associations; it does not establish novel resistance mechanisms, diagnostic accuracy, or patient-level clinical predictions.

International Journal of Molecular SciencesVol. 27(22)
Istanbul University-Cerrahpaşa (TR), Istanbul Provincial Health Directorate (TR), Istanbul University (TR)
Good health and well-being
Tuberculosis Research and Epidemiology
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon