Fabricio Alves Barbosa da Silva
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Fabricio Alves Barbosa da Silva has a total of 264 co-authors. Here are the top co-authors by number of shared publications:
- Hermes Senger
Department of Computer Science, Federal University of São Carlos, São Carlos, Brazil
16 shared publications - Nicolas Carels
Fundação Oswaldo Cruz
10 shared publications Marcelo Trindade dos Santos
Department of Computational Modeling, National Laboratory for Scientific Computing (LNCC/MCTI)
8 shared publications- Flávia Raquel Gonçalves Carneiro
Center of Technological Development in Health (CDTS), FIOCRUZ
7 shared publications - Maria Clicia Stelling de Carvalho
Department of Informatics and Computer Science, State University of Rio de Janeiro, Rio de Janeiro 20550-900, Brazil
7 shared publications - Domingos Alves
Ribeirao Preto Medical School, University of Sao Paulo, Ribeirao Preto, Brazil
6 shared publications
Latest Publications
- Open Access
The Tsallis index of the human cortical transcriptome is invariant between bipolar II disorder and control: a pre-registered, cross-platform null with the disease question relocated to correlation structure
Gene-expression fluctuations are heavy-tailed and resist description by additive (Boltzmann–Gibbs) statistics, motivating a non-extensive (Tsallis) account in which the index q summarizes the departure from Gaussianity. We ask two separable questions of the human cortical transcriptome in bipolar II disorder: whether the model-free value of q differs between cases and controls, and, if not, whether any disease signal instead resides in the correlation structure. Using a per-sample maximum-likelihood q -Gaussian estimator under a leakage-free, pre-registered protocol, we find the index concentrated near (GSE80655), (GSE12649), (GSE53987), with no case–control difference in any cohort. Per-sample model comparison favours the q -Gaussian over a Gaussian in 99–100% of samples (median ΔBIC ≪ 0), so q is a meaningful descriptor rather than an artefact of fitting. The BD–CTRL difference is null in every cohort. Pooling the three homogeneous cohorts, we test invariance by equivalence (TOST) rather than by non-rejection: the data establish equivalence at the bound |Δ q | ≥ 0.037, but do not reach the pre-registered bound of 0.03, which would require roughly twice the present sample. We report this shortfall explicitly: the null is bounded and informative, but the study is underpowered against its own pre-registered minimum effect of interest. A random-matrix test finds no between-group structural difference surviving label permutation. We argue that non-extensivity behaves as a conserved organizational property of the cortical transcriptome and that the disease question, for bipolar II, is relocated from the marginal index to the collective correlation modes and to dynamics (companion work). We report, and do not paper over, a failed cross-tissue scale anchor: a glioma RNA-seq reference is depth-confounded in a cohort-inconsistent way and therefore cannot license a same-value claim across tissues.
Open AccessThe Tsallis index of the human cortical transcriptome is invariant between bipolar II disorder and control: a pre-registered, cross-platform null with the disease question relocated to correlation structure
Gene-expression fluctuations are heavy-tailed and resist description by additive (Boltzmann–Gibbs) statistics, motivating a non-extensive (Tsallis) account in which the index q summarizes the departure from Gaussianity. We ask two separable questions of the human cortical transcriptome in bipolar II disorder: whether the model-free value of q differs between cases and controls, and, if not, whether any disease signal instead resides in the correlation structure. Using a per-sample maximum-likelihood q -Gaussian estimator under a leakage-free, pre-registered protocol, we find the index concentrated near (GSE80655), (GSE12649), (GSE53987), with no case–control difference in any cohort. Per-sample model comparison favours the q -Gaussian over a Gaussian in 99–100% of samples (median ΔBIC ≪ 0), so q is a meaningful descriptor rather than an artefact of fitting. The BD–CTRL difference is null in every cohort. Pooling the three homogeneous cohorts, we test invariance by equivalence (TOST) rather than by non-rejection: the data establish equivalence at the bound |Δ q | ≥ 0.037, but do not reach the pre-registered bound of 0.03, which would require roughly twice the present sample. We report this shortfall explicitly: the null is bounded and informative, but the study is underpowered against its own pre-registered minimum effect of interest. A random-matrix test finds no between-group structural difference surviving label permutation. We argue that non-extensivity behaves as a conserved organizational property of the cortical transcriptome and that the disease question, for bipolar II, is relocated from the marginal index to the collective correlation modes and to dynamics (companion work). We report, and do not paper over, a failed cross-tissue scale anchor: a glioma RNA-seq reference is depth-confounded in a cohort-inconsistent way and therefore cannot license a same-value claim across tissues.
- Open Access
- 10 July 2026
- Published by MDPI AG in Preprints.org
Precision Therapeutics in Triple-Negative Breast Cancer: Boolean Network Control and Pharmacological Mapping of Individualized Models
Triple-negative Breast cancer is a highly heterogeneous tumor subtype lacking universal therapeutic targets, a factor that frequently renders standard pharmacological approaches ineffective. To address the urgent clinical need for precision strategies, we developed a computational framework centered on personalization. Utilizing specific gene expression data, we initially constructed and calibrated 41 individualized regulatory networks, thereby capturing the precise molecular signature of each patient. Then, we applied a hybrid approach to these models—integrating Boolean logical determinism with the exploratory optimization of the combinatorial search space via a Genetic Algorithm—and identified the minimal sets of driver nodes capable of irreversibly driving the tumor cell toward apoptosis. Subsequently, a rigorous optimization process translated these targets into actionable clinical options by mapping them to optimal combinations of oncological drugs under stringent therapeutic constraints. The apoptotic transition was successfully achieved in 92.7% of the samples. Results indicate that effective tumor reprogramming relies on manipulating critical nodes that orchestrate the network's logical dynamics. By minimizing unnecessary perturbations to the overall signaling architecture, this methodology demonstrates the computational feasibility of identifying robust, non-obvious therapeutic targets through Boolean modeling. Ultimately, it offers an adaptive and scalable platform for the rational design of minimally disruptive, patient-specific combinatorial therapies.
- 10 July 2026
- Published by MDPI AG in Preprints.org
Open AccessPrecision Therapeutics in Triple-Negative Breast Cancer: Boolean Network Control and Pharmacological Mapping of Individualized Models
Triple-negative Breast cancer is a highly heterogeneous tumor subtype lacking universal therapeutic targets, a factor that frequently renders standard pharmacological approaches ineffective. To address the urgent clinical need for precision strategies, we developed a computational framework centered on personalization. Utilizing specific gene expression data, we initially constructed and calibrated 41 individualized regulatory networks, thereby capturing the precise molecular signature of each patient. Then, we applied a hybrid approach to these models—integrating Boolean logical determinism with the exploratory optimization of the combinatorial search space via a Genetic Algorithm—and identified the minimal sets of driver nodes capable of irreversibly driving the tumor cell toward apoptosis. Subsequently, a rigorous optimization process translated these targets into actionable clinical options by mapping them to optimal combinations of oncological drugs under stringent therapeutic constraints. The apoptotic transition was successfully achieved in 92.7% of the samples. Results indicate that effective tumor reprogramming relies on manipulating critical nodes that orchestrate the network's logical dynamics. By minimizing unnecessary perturbations to the overall signaling architecture, this methodology demonstrates the computational feasibility of identifying robust, non-obvious therapeutic targets through Boolean modeling. Ultimately, it offers an adaptive and scalable platform for the rational design of minimally disruptive, patient-specific combinatorial therapies.
- Open Access
- 15 June 2026
- Published by MDPI AG in International Journal of Molecular Sciences
- Vol. 27 (12) , 5385
Hybrid Computational Modeling with Multi-Level Validation Identifies TK1–VIM as a Robust Therapeutic Pair in Triple-Negative Breast Cancer
Triple-negative breast cancer (TNBC) lacks effective molecular targets, leading to poor prognosis. Previous computational methods to identify targets have suffered from low druggability, high complexity, and lack of robust validation. We propose a hybrid methodology combining Boolean network modeling with semidefinite programming (SDP) to analyze a TNBC cell line network. The resulting therapeutic pair underwent a multi-level validation framework, including Boolean simulations, statistical uncertainty quantification (bootstrap), sensitivity analysis, and orthogonal computational support from AlphaGenome, a deep learning model from Google DeepMind. Our analysis identified TK1 and VIM as a computationally robust therapeutic pair. Dual inhibition achieved 99.03% similarity to the apoptotic state with a 95% confidence interval of [98.79%, 99.26%], and was statistically superior to alternative pairs (). The selection remained optimal across all tested model parameters, demonstrating high robustness. Importantly, the pair has full druggability because both targets have available specific inhibitors. Orthogonal computational evidence from AlphaGenome, stratified by mammary compartment, indicated that both targets exhibit moderate baseline expression in normal mammary epithelium (TK1 = 0.159, VIM = 0.143 in normalized RNA-seq units; n = 13 tracks per gene), with VIM showing a 2.2-fold higher expression in mammary stroma than in epithelium—a gradient consistent with its established role as a mesenchymal marker. Promoter-variant proxy analysis indicated near-zero transcriptomic perturbation upon simulated inhibition of either target in normal mammary epithelium (mean ), supporting a favorable therapeutic window. Our methodology identified TK1–VIM as a computationally robust, druggable therapeutic candidate pair with biologically plausible mechanism of action. Gene-variability analysis identified TK1 and VIM as the highest-scoring candidates, with SDP optimization providing complementary, independent confirmation of this selection. This work provides a computationally grounded candidate strategy and a rigorous methodological benchmark for computational drug target identification; experimental validation remains an essential next step before clinical translation.- 15 June 2026
- Published by MDPI AG in International Journal of Molecular Sciences
- Vol. 27 (12) , 5385
Open AccessHybrid Computational Modeling with Multi-Level Validation Identifies TK1–VIM as a Robust Therapeutic Pair in Triple-Negative Breast Cancer
Triple-negative breast cancer (TNBC) lacks effective molecular targets, leading to poor prognosis. Previous computational methods to identify targets have suffered from low druggability, high complexity, and lack of robust validation. We propose a hybrid methodology combining Boolean network modeling with semidefinite programming (SDP) to analyze a TNBC cell line network. The resulting therapeutic pair underwent a multi-level validation framework, including Boolean simulations, statistical uncertainty quantification (bootstrap), sensitivity analysis, and orthogonal computational support from AlphaGenome, a deep learning model from Google DeepMind. Our analysis identified TK1 and VIM as a computationally robust therapeutic pair. Dual inhibition achieved 99.03% similarity to the apoptotic state with a 95% confidence interval of [98.79%, 99.26%], and was statistically superior to alternative pairs (). The selection remained optimal across all tested model parameters, demonstrating high robustness. Importantly, the pair has full druggability because both targets have available specific inhibitors. Orthogonal computational evidence from AlphaGenome, stratified by mammary compartment, indicated that both targets exhibit moderate baseline expression in normal mammary epithelium (TK1 = 0.159, VIM = 0.143 in normalized RNA-seq units; n = 13 tracks per gene), with VIM showing a 2.2-fold higher expression in mammary stroma than in epithelium—a gradient consistent with its established role as a mesenchymal marker. Promoter-variant proxy analysis indicated near-zero transcriptomic perturbation upon simulated inhibition of either target in normal mammary epithelium (mean ), supporting a favorable therapeutic window. Our methodology identified TK1–VIM as a computationally robust, druggable therapeutic candidate pair with biologically plausible mechanism of action. Gene-variability analysis identified TK1 and VIM as the highest-scoring candidates, with SDP optimization providing complementary, independent confirmation of this selection. This work provides a computationally grounded candidate strategy and a rigorous methodological benchmark for computational drug target identification; experimental validation remains an essential next step before clinical translation. - Open Access
- 28 May 2026
- Published by Frontiers Media SA in Frontiers in Cellular and Infection Microbiology
- Vol. 16, 1859508
Harnessing interpretable deep learning to predict resistance in Klebsiella pneumoniae
Introduction: Antimicrobial resistance poses a growing global health threat, complicating therapeutic management and increasing morbidity and mortality. Deep learning methods have emerged as effective tools for bacterial profiling using omics data, particularly in predicting antimicrobial susceptibility from genomic information. The present study focuses on identifying genomic signatures linked to resistance mechanisms using a deep learning architecture. Methods: DeepMDC, a deep learning architecture for bacterial profiling using whole-genome data, is introduced. Due to the high cost and ambiguity of precise gene- or mutation-level annotation, phenotypic classification is formulated as a Multiple Instance Learning (MIL) problem, in which each genome is represented as a bag of instances with a single associated label. The architecture centers on a modern Hopfield network that processes all open reading frames (ORFs), including small ORFs, derived from genomic data. Interpretability is achieved through attention mechanisms, which facilitate biological insight and support hypothesis generation. Results: The model was evaluated against Klebsiella pneumoniae and four clinically relevant antibiotics (meropenem, cefepime, ceftazidime, and gentamicin), achieving strong performance across multiple evaluation metrics. Discussion: Notably, genes associated with resistance consistently received high attention scores during inference, validating the architecture and potentially generating new hypotheses.
- 28 May 2026
- Published by Frontiers Media SA in Frontiers in Cellular and Infection Microbiology
- Vol. 16, 1859508
Open AccessHarnessing interpretable deep learning to predict resistance in Klebsiella pneumoniae
Introduction: Antimicrobial resistance poses a growing global health threat, complicating therapeutic management and increasing morbidity and mortality. Deep learning methods have emerged as effective tools for bacterial profiling using omics data, particularly in predicting antimicrobial susceptibility from genomic information. The present study focuses on identifying genomic signatures linked to resistance mechanisms using a deep learning architecture. Methods: DeepMDC, a deep learning architecture for bacterial profiling using whole-genome data, is introduced. Due to the high cost and ambiguity of precise gene- or mutation-level annotation, phenotypic classification is formulated as a Multiple Instance Learning (MIL) problem, in which each genome is represented as a bag of instances with a single associated label. The architecture centers on a modern Hopfield network that processes all open reading frames (ORFs), including small ORFs, derived from genomic data. Interpretability is achieved through attention mechanisms, which facilitate biological insight and support hypothesis generation. Results: The model was evaluated against Klebsiella pneumoniae and four clinically relevant antibiotics (meropenem, cefepime, ceftazidime, and gentamicin), achieving strong performance across multiple evaluation metrics. Discussion: Notably, genes associated with resistance consistently received high attention scores during inference, validating the architecture and potentially generating new hypotheses.
- Open Access
Tsallis-Gated Autoencoder: A Nonextensive Physics-Informed Approach for Unsupervised Anomaly Detection in Glioblastoma Multiforme RNA-seq Data
Glioblastoma multiforme (GBM) is characterised by profound genomic heterogeneity and heavy-tailed gene-expression distributions that challenge conventional machine-learning methods. We introduce the Tsallis-Gated Autoencoder (Tsallis-GAE), a physics-informed architecture that replaces classical softmax attention with a learnable Tsallis q -softmax followed by mean-field smoothing iterations, motivated by recent work on curved statistical manifolds and dense associative networks. Trained on the full TCGA-GBM RNA-seq cohort (391 samples, top 2,000 high-variance genes) under a rigorous 80/20 hold-out protocol, the Tsallis-GAE achieves a mean AUC-ROC of 0.977 ± 0.002 across five independent seeds, compared to 0.906 ± 0.003 for a matched-capacity Vanilla autoencoder trained under the identical protocol. The matched-capacity Vanilla autoencoder is statistically indistinguishable from a LocalOutlierFactor baseline (AUC 0.906 vs 0.906), confirming that the +0.07 AUC gain over the Vanilla AE stems from the gated attention architecture rather than from the use of a neural network per se. A fixed- q Softmax-AE ablation ( q ≡1 by construction) achieves AUC 0.976 ±0.001, only +0.001 below the Tsallis-GAE (DeLong p = 0.44); the physically meaningful contribution of the learnable q is its spontaneous convergence to the non-extensive regime described below. The three attention blocks each carry an independent learnable entropic index q ; across 5 seeds ×3 blocks = 15 measurements, q converges spontaneously to 1.554± 0.019, strictly bounded away from the Boltzmann–Gibbs limit q = 1 and in the moderate non-extensivity regime characteristic of complex biological systems. Cross-detector validation against OneClassSVM and LocalOutlier-Factor pseudo-labels yields Tsallis-GAE AUCs of 0.998 and 0.992 respectively, indicating that the learned representation captures anomaly structure intrinsic to the data rather than the decision boundary of any single labeling heuristic. We declare that DeLong’s paired test on the present test-set size ( n = 79) does not certify the +0.07 AUC gap as formally significant ( p ≈ 0.26); a 5-fold cross-validation over the full cohort, which would supply the needed statistical power, is left to future work. The source code is available upon reasonable request to the corresponding author.
Open AccessTsallis-Gated Autoencoder: A Nonextensive Physics-Informed Approach for Unsupervised Anomaly Detection in Glioblastoma Multiforme RNA-seq Data
Glioblastoma multiforme (GBM) is characterised by profound genomic heterogeneity and heavy-tailed gene-expression distributions that challenge conventional machine-learning methods. We introduce the Tsallis-Gated Autoencoder (Tsallis-GAE), a physics-informed architecture that replaces classical softmax attention with a learnable Tsallis q -softmax followed by mean-field smoothing iterations, motivated by recent work on curved statistical manifolds and dense associative networks. Trained on the full TCGA-GBM RNA-seq cohort (391 samples, top 2,000 high-variance genes) under a rigorous 80/20 hold-out protocol, the Tsallis-GAE achieves a mean AUC-ROC of 0.977 ± 0.002 across five independent seeds, compared to 0.906 ± 0.003 for a matched-capacity Vanilla autoencoder trained under the identical protocol. The matched-capacity Vanilla autoencoder is statistically indistinguishable from a LocalOutlierFactor baseline (AUC 0.906 vs 0.906), confirming that the +0.07 AUC gain over the Vanilla AE stems from the gated attention architecture rather than from the use of a neural network per se. A fixed- q Softmax-AE ablation ( q ≡1 by construction) achieves AUC 0.976 ±0.001, only +0.001 below the Tsallis-GAE (DeLong p = 0.44); the physically meaningful contribution of the learnable q is its spontaneous convergence to the non-extensive regime described below. The three attention blocks each carry an independent learnable entropic index q ; across 5 seeds ×3 blocks = 15 measurements, q converges spontaneously to 1.554± 0.019, strictly bounded away from the Boltzmann–Gibbs limit q = 1 and in the moderate non-extensivity regime characteristic of complex biological systems. Cross-detector validation against OneClassSVM and LocalOutlier-Factor pseudo-labels yields Tsallis-GAE AUCs of 0.998 and 0.992 respectively, indicating that the learned representation captures anomaly structure intrinsic to the data rather than the decision boundary of any single labeling heuristic. We declare that DeLong’s paired test on the present test-set size ( n = 79) does not certify the +0.07 AUC gap as formally significant ( p ≈ 0.26); a 5-fold cross-validation over the full cohort, which would supply the needed statistical power, is left to future work. The source code is available upon reasonable request to the corresponding author.
- Open Access
- 11 May 2026
- Published by MDPI AG in Preprints.org
Hybrid Computational Modeling with Multi-Level Validation Identifies TK1-VIM as a Robust Therapeutic Pair in Triple-Negative Breast Cancer
Triple-negative breast cancer (TNBC) lacks effective molecular targets, leading to poor prognosis. Previous computational methods to identify targets have suffered from low druggability, high complexity, and lack of robust validation. We propose a hybrid methodology combining Boolean network modeling with semidefinite programming (SDP) to analyze a TNBC cell line network. The resulting therapeutic pair underwent a multi-level validation framework, including Boolean simulations, statistical uncertainty quantification (bootstrap), sensitivity analysis, and independent verification by AlphaGenome v2, a deep learning model from Google DeepMind. Our analysis identified TK1 and VIM as a robust therapeutic pair. Dual inhibition achieved 99.03% similarity to the apoptotic state with a 95% confidence interval of [98.79%, 99.26%], and was statistically superior to alternative pairs (p < 0.001). The selection remained optimal across all tested model parameters, demonstrating high robustness. Importantly, the pair has full druggability because both targets have available specific inhibitors. AlphaGenome v2 validation in normal mammary tissue revealed that TK1 exhibits moderate expression while VIM shows low baseline expression. This differential pattern, combined with strong VIM upregulation in the mesenchymal-like TNBC phenotype, supports the synergistic mechanism of the dual-target strategy. Our methodology identified TK1-VIM as a high-confidence and druggable therapeutic pair for TNBC with strong biological plausibility. This work provides a clinically actionable strategy and establishes a new benchmark for computational rigor in drug target identification.
- 11 May 2026
- Published by MDPI AG in Preprints.org
Open AccessHybrid Computational Modeling with Multi-Level Validation Identifies TK1-VIM as a Robust Therapeutic Pair in Triple-Negative Breast Cancer
Triple-negative breast cancer (TNBC) lacks effective molecular targets, leading to poor prognosis. Previous computational methods to identify targets have suffered from low druggability, high complexity, and lack of robust validation. We propose a hybrid methodology combining Boolean network modeling with semidefinite programming (SDP) to analyze a TNBC cell line network. The resulting therapeutic pair underwent a multi-level validation framework, including Boolean simulations, statistical uncertainty quantification (bootstrap), sensitivity analysis, and independent verification by AlphaGenome v2, a deep learning model from Google DeepMind. Our analysis identified TK1 and VIM as a robust therapeutic pair. Dual inhibition achieved 99.03% similarity to the apoptotic state with a 95% confidence interval of [98.79%, 99.26%], and was statistically superior to alternative pairs (p < 0.001). The selection remained optimal across all tested model parameters, demonstrating high robustness. Importantly, the pair has full druggability because both targets have available specific inhibitors. AlphaGenome v2 validation in normal mammary tissue revealed that TK1 exhibits moderate expression while VIM shows low baseline expression. This differential pattern, combined with strong VIM upregulation in the mesenchymal-like TNBC phenotype, supports the synergistic mechanism of the dual-target strategy. Our methodology identified TK1-VIM as a high-confidence and druggable therapeutic pair for TNBC with strong biological plausibility. This work provides a clinically actionable strategy and establishes a new benchmark for computational rigor in drug target identification.
- Open Access
Identification of Therapeutic Gene Targets in Triple-Negative Breast Cancer: A Hybrid Approach Integrating Semidefinite Programming, Boolean Simulation, and Druggability Analysis
Triple-negative breast cancer (TNBC) represents a significant clinical challenge due to the absence of well-established molecular targets and resistance to conventional therapies. Identifying synergistic combinations of therapeutic targets requires computational approaches that integrate topological network analysis, experimental validation, and clinical applicability. This work proposes a hybrid methodology that combines Semidefinite Programming (SDP) for identification of critical nodes in gene regulatory networks, Boolean simulation for quantification of perturbation efficacy, and systematic literature validation with focus on druggability and synergy. We constructed a core regulatory network of 13 genes representing key nodes in STAT3, PI3K/AKT, and p53 signaling pathways, which are frequently deregulated in TNBC. We applied three complementary SDP formulations (Max-Cut, Influence Maximization, and Spectral Clustering) to identify candidate targets, followed by stochastic Boolean simulation for calculation of the Therapeutic Index (TI). We integrated a practical applicability scoring system that considers druggability, clinical evidence, specificity, and TNBC validation. Our analysis identified the combination STAT3 + BCL2 as the most promising therapeutic pair, with an applicability score of 0.905 and average druggability of 0.85. This finding is strongly supported by extensive experimental evidence demonstrating that STAT3 directly regulates BCL2 transcription in breast cancer cells. We demonstrate a clear methodological evolution from previous approaches: from 5 genes (Tilli et al., 2016, druggability 0.32) to 3 genes ([9], druggability 0.40) and finally to 2 genes (our work, druggability 0.85), representing a 166% increase in druggability and 60% reduction in complexity. The proposed methodology offers a systematic and reproducible framework for prioritization of therapeutic targets with focus on clinical applicability, contributing to rational development of combination therapies in TNBC.
Open AccessIdentification of Therapeutic Gene Targets in Triple-Negative Breast Cancer: A Hybrid Approach Integrating Semidefinite Programming, Boolean Simulation, and Druggability Analysis
Triple-negative breast cancer (TNBC) represents a significant clinical challenge due to the absence of well-established molecular targets and resistance to conventional therapies. Identifying synergistic combinations of therapeutic targets requires computational approaches that integrate topological network analysis, experimental validation, and clinical applicability. This work proposes a hybrid methodology that combines Semidefinite Programming (SDP) for identification of critical nodes in gene regulatory networks, Boolean simulation for quantification of perturbation efficacy, and systematic literature validation with focus on druggability and synergy. We constructed a core regulatory network of 13 genes representing key nodes in STAT3, PI3K/AKT, and p53 signaling pathways, which are frequently deregulated in TNBC. We applied three complementary SDP formulations (Max-Cut, Influence Maximization, and Spectral Clustering) to identify candidate targets, followed by stochastic Boolean simulation for calculation of the Therapeutic Index (TI). We integrated a practical applicability scoring system that considers druggability, clinical evidence, specificity, and TNBC validation. Our analysis identified the combination STAT3 + BCL2 as the most promising therapeutic pair, with an applicability score of 0.905 and average druggability of 0.85. This finding is strongly supported by extensive experimental evidence demonstrating that STAT3 directly regulates BCL2 transcription in breast cancer cells. We demonstrate a clear methodological evolution from previous approaches: from 5 genes (Tilli et al., 2016, druggability 0.32) to 3 genes ([9], druggability 0.40) and finally to 2 genes (our work, druggability 0.85), representing a 166% increase in druggability and 60% reduction in complexity. The proposed methodology offers a systematic and reproducible framework for prioritization of therapeutic targets with focus on clinical applicability, contributing to rational development of combination therapies in TNBC.
- Open Access
RNA-seq analysis reveals transcriptional disruption induced by combined cisplatin and TiO□ nanoparticles in Drosophila melanogaster
Cisplatin, the first metal-based chemotherapeutic drug, remains widely used despite its toxicity and the emergence of drug resistance. Combining cisplatin with nanoparticles has been proposed to improve its therapeutic profile, although most studies rely on in vitro models. Here, we investigated the in vivo transcriptional effects of cisplatin (50 μg/ml) and titanium dioxide nanoparticles (TiO□ NPs; 50 μg/ml), alone and in combination, in Drosophila melanogaster . Using RNA-seq and bioinformatic analyses, we assessed differential gene expression across treatments. Flies exposed to cisplatin or TiO□ NPs alone exhibited modulation of genes associated with xenobiotic metabolism and detoxification. In contrast, combined exposure resulted in a markedly reduced number of upregulated differentially expressed genes (DEGs): CG10013, aqz, CG5568 , and CG3213 , which are related to cell division and DNA/RNA metabolism. Also, we observed that the combined exposure promotes the downregulation of DEGs involved in xenobiotic metabolism, detoxification, collapse of the innate immune system, and reproductive impairment and fertility. Mortality rates were not significantly affected in any group. These results suggest that co-exposure to cisplatin and TiO□ nanoparticles induces a global transcriptional suppression, potentially impairing essential cellular processes.
Open AccessRNA-seq analysis reveals transcriptional disruption induced by combined cisplatin and TiO□ nanoparticles in Drosophila melanogaster
Cisplatin, the first metal-based chemotherapeutic drug, remains widely used despite its toxicity and the emergence of drug resistance. Combining cisplatin with nanoparticles has been proposed to improve its therapeutic profile, although most studies rely on in vitro models. Here, we investigated the in vivo transcriptional effects of cisplatin (50 μg/ml) and titanium dioxide nanoparticles (TiO□ NPs; 50 μg/ml), alone and in combination, in Drosophila melanogaster . Using RNA-seq and bioinformatic analyses, we assessed differential gene expression across treatments. Flies exposed to cisplatin or TiO□ NPs alone exhibited modulation of genes associated with xenobiotic metabolism and detoxification. In contrast, combined exposure resulted in a markedly reduced number of upregulated differentially expressed genes (DEGs): CG10013, aqz, CG5568 , and CG3213 , which are related to cell division and DNA/RNA metabolism. Also, we observed that the combined exposure promotes the downregulation of DEGs involved in xenobiotic metabolism, detoxification, collapse of the innate immune system, and reproductive impairment and fertility. Mortality rates were not significantly affected in any group. These results suggest that co-exposure to cisplatin and TiO□ nanoparticles induces a global transcriptional suppression, potentially impairing essential cellular processes.
- Open Access
Harnessing Interpretable Deep Learning to Predict Resistance in Klebsiella pneumoniae
Antimicrobial resistance constitutes an escalating global health threat, complicating therapeutic management and increasing morbidity and mortality. Deep learning approaches have emerged as promising tools for bacterial profiling based on omics data, particularly for predicting antimicrobial susceptibility from genomic information. This task relies on identifying genomic signatures associated with resistance mechanisms. Here, DeepMDC is introduced as a deep learning architecture designed for bacterial profiling using whole-genome data. Given that precise annotation at the gene or mutation level is often costly and ambiguous, phenotypic classification is formulated as a Multiple Instance Learning (MIL) problem, in which each genome is represented as a bag of instances with a single associated label. The core of DeepMDC is a Modern Hopfield Network that processes all open reading frames (ORFs), including small ones, derived from genomic data. A key feature of the architecture is its interpretability, enabled by attention mechanisms that facilitate biological insight and hypothesis generation. The model was evaluated against Klebsiella pneumoniae and four clinically relevant antibiotics (meropenem, cefepime, ceftazidime, and gentamicin), achieving strong performance in several metrics. Notably, genes associated with resistance consistently received high attention scores during inference, which validates the architecture and eventually may generate new hypotheses.
Open AccessHarnessing Interpretable Deep Learning to Predict Resistance in Klebsiella pneumoniae
Antimicrobial resistance constitutes an escalating global health threat, complicating therapeutic management and increasing morbidity and mortality. Deep learning approaches have emerged as promising tools for bacterial profiling based on omics data, particularly for predicting antimicrobial susceptibility from genomic information. This task relies on identifying genomic signatures associated with resistance mechanisms. Here, DeepMDC is introduced as a deep learning architecture designed for bacterial profiling using whole-genome data. Given that precise annotation at the gene or mutation level is often costly and ambiguous, phenotypic classification is formulated as a Multiple Instance Learning (MIL) problem, in which each genome is represented as a bag of instances with a single associated label. The core of DeepMDC is a Modern Hopfield Network that processes all open reading frames (ORFs), including small ones, derived from genomic data. A key feature of the architecture is its interpretability, enabled by attention mechanisms that facilitate biological insight and hypothesis generation. The model was evaluated against Klebsiella pneumoniae and four clinically relevant antibiotics (meropenem, cefepime, ceftazidime, and gentamicin), achieving strong performance in several metrics. Notably, genes associated with resistance consistently received high attention scores during inference, which validates the architecture and eventually may generate new hypotheses.
- Open Access
Proteomic profiling of skeletal muscle ribosomes from higher versus lower responders to 10 weeks of resistance training
Ribosome biogenesis is a key driver of resistance training (RT)-induced skeletal muscle hypertrophy in humans. However, high resolution insight into RT-induced compositional alterations in ribosomes remain unexplored. Therefore, the purpose of this study was twofold: 1) develop a protocol sufficient to enrich and examine the ribosomal proteome from a small amount of human muscle tissue, and 2) determine if ribosomal protein composition differs between higher and lower responders to RT. Fourteen participants completed 10 weeks of RT (23 sessions) and were stratified into higher (n=7) and lower (n=7) responders based on changes in vastus lateralis muscle cross-sectional area (VL mCSA) and mean myofiber cross-sectional area (fCSA) from baseline (PRE) to after the 23 rd RT session (POST). Participants then performed a twenty fourth RT session and biopsies were collected 24 hours post-RT (POST-24h) to examine acute bout (POST to POST-24h) ribosomal proteome alterations to RT. Ribosome enrichment and analysis included ultracentrifugation (100,000 g , 3h, 2°C) through 20% sucrose gradients followed by shotgun proteomics to quantify ribosomal protein composition. Our protocol produced exceptional ribosome enrichment, with 74 distinct ribosomal proteins being detected (92% coverage of the putative 80 cytosolic ribosomal proteins) as well as ∼164-fold and ∼71-fold increases in large (RPL) and small (RPS) protein subunit abundances, respectively, compared to conventional tissue homogenization. Despite robust phenotypic differences in hypertrophy between responder groups to the 10-week RT protocol (mCSA: +31% versus +3%; fCSA: +29% versus -2%, P0.028 for all comparisons). In conclusion, we present a highly effective ribosome enrichment protocol requiring minimal human muscle tissue. Though preliminary analyses indicate that ribosomal protein composition was similar between hypertrophic phenotypes in response to an acute RT session after a 10-week RT intervention, future research leveraging our techniques with more sampling time points are needed to provide more definitive conclusions.
Open AccessProteomic profiling of skeletal muscle ribosomes from higher versus lower responders to 10 weeks of resistance training
Ribosome biogenesis is a key driver of resistance training (RT)-induced skeletal muscle hypertrophy in humans. However, high resolution insight into RT-induced compositional alterations in ribosomes remain unexplored. Therefore, the purpose of this study was twofold: 1) develop a protocol sufficient to enrich and examine the ribosomal proteome from a small amount of human muscle tissue, and 2) determine if ribosomal protein composition differs between higher and lower responders to RT. Fourteen participants completed 10 weeks of RT (23 sessions) and were stratified into higher (n=7) and lower (n=7) responders based on changes in vastus lateralis muscle cross-sectional area (VL mCSA) and mean myofiber cross-sectional area (fCSA) from baseline (PRE) to after the 23 rd RT session (POST). Participants then performed a twenty fourth RT session and biopsies were collected 24 hours post-RT (POST-24h) to examine acute bout (POST to POST-24h) ribosomal proteome alterations to RT. Ribosome enrichment and analysis included ultracentrifugation (100,000 g , 3h, 2°C) through 20% sucrose gradients followed by shotgun proteomics to quantify ribosomal protein composition. Our protocol produced exceptional ribosome enrichment, with 74 distinct ribosomal proteins being detected (92% coverage of the putative 80 cytosolic ribosomal proteins) as well as ∼164-fold and ∼71-fold increases in large (RPL) and small (RPS) protein subunit abundances, respectively, compared to conventional tissue homogenization. Despite robust phenotypic differences in hypertrophy between responder groups to the 10-week RT protocol (mCSA: +31% versus +3%; fCSA: +29% versus -2%, P0.028 for all comparisons). In conclusion, we present a highly effective ribosome enrichment protocol requiring minimal human muscle tissue. Though preliminary analyses indicate that ribosomal protein composition was similar between hypertrophic phenotypes in response to an acute RT session after a 10-week RT intervention, future research leveraging our techniques with more sampling time points are needed to provide more definitive conclusions.