MPFFPSDC: A multi-pooling feature fusion model for predicting synergistic drug combinations.

<div>Abstract<p>Combination drug therapy is a widely used paradigm for managing numerous human malignancies. In cancer treatment, additive and/or synergistic drug combinations can convert weakly efficacious monotherapies into regimens that produce robust antitumor activity. This can be explained in part through pathway interdependencies that are critical for cancer cell proliferation and survival. However, identification of the various interdependencies is difficult due to the complex molecular circuitry that underlies tumor development and progression. Here, we present a high-throughput platform that allows for an unbiased identification of synergistic and efficacious drug combinations. In a screen of 22,737 experiments of 583 doublet combinations in 39 diverse cancer cell lines using a 4 by 4 dosing regimen, both well-known and novel synergistic and efficacious combinations were identified. Here, we present an example of one such novel combination, a Wee1 inhibitor (AZD1775) and an mTOR inhibitor (ridaforolimus), and demonstrate that the combination potently and synergistically inhibits cancer cell growth <i>in vitro</i> and <i>in vivo</i>. This approach has identified novel combinations that would be difficult to reliably predict based purely on our current understanding of cancer cell biology. <i>Mol Cancer Ther; 15(6); 1155–62. ©2016 AACR</i>.</p></div>

Drug combination therapy often outperforms monotherapy in cancer treatment, but the vast number of available drugs makes manual screening for synergistic combinations costly. Computational methods, especially deep learning, can reduce the search space by predicting likely synergistic drug combinations. Recent studies have improved drug synergy prediction by modeling associations among different biological entities, but drug-drug interactions have not been fully leveraged in this scenario, which motivated the work presented in this paper. This paper proposes a deep learning method named HGTSynergy to predict synergistic drug combinations, which employs a heterogeneous graph attention network and a tailored task to capture complex latent patterns in the drug network as prior knowledge. The learned knowledge is then transferred through a transfer learning framework to the downstream task of predicting drug synergy scores, effectively enhancing predictive performance. A five-fold nested cross-validation is employed to train HGTSynergy. In the synergy regression task, HGTSynergy outperforms seven deep learning methods, achieving a mean squared error of 222.83, root mean squared error of 14.91, and Pearson correlation coefficient of 0.75. For the synergy classification task, it also surpasses other methods with an area under the receiver operating characteristic curve of 0.90, area under the precision-recall curve of 0.63, accuracy of 0.94, precision of 0.72, and Cohen's Kappa of 0.52. The ablation study verifies that the heterogeneous graph attention network and the transfer learning framework both have a positive effect on prediction performance. Moreover, a series of analyses demonstrates that the proposed method exhibits strong generalization performance and interpretability. The case study further validates its consistency with prior research. This study suggests that drug synergy prediction can be improved by comprehensively modeling diverse drug-drug interaction types and leveraging transfer learning to extract prior knowledge from them. The ability of HGTSynergy to discover new anticancer synergistic drug combinations outperforms other state-of-the-art methods. HGTSynergy promises to be a powerful tool to pre-screen anticancer synergistic drug combinations.

Drug combination therapies can improve drug efficacy, reduce drug dosage, and overcome drug resistance in cancer treatments. Current research strategies to determine which drug combinations have a synergistic effect rely mainly on clinical or empirical experience and screening predefined pools of drugs. Given the number of possible drug combinations, the speed, and scope to find new drug combinations are very limited using these methods. Due to the exponential growth in the number of drug combinations, it is difficult to test all possible combinations in the lab. There are several large-scale public genomic and phenotypic resources that provide data from single drug-treated cells as well as data from small molecule treated cells. These databases provide a wealth of information regarding cellular responses to drugs and offer an opportunity to overcome the limitations of the current methods. Developing a new advanced data processing and analysis strategy is imperative and a computational prediction algorithm is highly desirable. In this paper, we developed a computational algorithm for the enrichment of synergistic drug combinations using gene regulatory network knowledge and an operational module unit (OMU) system which we generate from single drug genomic and phenotypic data. As a proof of principle, we applied the pipeline to a group of anticancer drugs and demonstrate how the algorithm could help researchers efficiently find possible synergistic drug combinations using single drug data to evaluate all possible drug pairs.

Neuroblastoma is the most common tumor in infants younger than 1 year of age and one of the most common solid tumors in childhood. The disease is highly variable: while spontaneous tumor regression occurs more often than in any other cancer type, patients with more severe variant face therapy-resistant recurrence. Therefore we have set out to develop new therapeutic approaches to the disease with a particular focus on aggressive subtypes. Using a high-throughput systems level approach we have identified a novel synergistic combination of approved drugs and compounds in clinical trials. Our initial screen assessed the potency of a panel of clinically applicable drugs by proliferation assays versus a defined panel of isogenic cell lines that are representative of the different neuroblastoma variants. We identified potential drugs combinations by three-dimensional dose response matrices and adopted Bliss values as a quantitative measure of the synergy. The striking synergy was observed in SH-SY5Y cell line using combinations of EGFR/HER2 inhibitors with the apoptosis inducing agents. We were able to validate the observation using colony formation assay. In contrast to single drug treatment, the combination completely abolished the clonogenic growth of neuroblastoma cells, showing the most tangible evidence of the synergy. In order to evaluate our results in more aggressive disease subtypes, we tested our combination in SH-SY5Y cell line with transiently overexpressed N-Myc and in N-Myc-amplified IMR5-75 cell line. Notably, even in this setting we could recapitulate our findings, suggesting that this combination of drugs is highly potent in the chemo-resistant tumor subtypes as well. The same effect was observed when we transiently overexpressed TrkA in SH-SY5Y, mimicking yet another neuroblastoma variant. Pursuing the exact mechanism of the synergy, we have tested the combination in breast cancer cells. To our surprise, the synergistic effect was still present in HER2 negative cell line, but completely absent in the HER2 positive one. We also exchanged the dual EGFR/HER2 inhibitor drug with more potent EGFR inhibitors to conclude that a combination seems to be specific to a particular set of compounds and not necessarily to their main target spectra. Our synergistic combination is highly potent in a large number of neuroblastoma conditions. Strong synergistic effect can be reached at doses much lower than required for the single drug treatment. Animal models that we are currently testing will enable quick translation of our data in clinics. Citation Format: Branka Radic, Uwe Rix, Stefan Kubicek, Giulio Superti-Furga. Probing network fragilities in neuroblastoma by synergistic drug combinations. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 5038. doi:10.1158/1538-7445.AM2013-5038

Drug combination therapies are promising clinical treatments for curing patients. However, efficiently identifying valid drug combinations remains challenging because the number of available drugs has increased rapidly. In this study, we proposed a deep learning model called the Dual Feature Fusion Network for Drug–Drug Synergy prediction (DFFNDDS) that utilizes a fine-tuned pretrained language model and dual feature fusion mechanism to predict synergistic drug combinations. The dual feature fusion mechanism fuses the drug features and cell line features at the bit-wise level and the vector-wise level. We demonstrated that DFFNDDS outperforms competitive methods and can serve as a reliable tool for identifying synergistic drug combinations.

Identification of effective drug combinations inhibiting multiple target genes that are essential for cancer cell survival is seen as an effective strategy to overcome drug resistance. Cellular pathways are often redundant, and network rewiring in the face of an assault promotes the emergence of resistant cells that contribute to clinical relapse. However, the number of possible drug-target combinations to investigate by conducting experiments is exponentially large and impractical, therefore warranting the need for systematic approaches to prioritize the most effective drug combinations. Here, we aimed to integrate large-scale drug screening, drug-target affinity knowledge and public RNAi screening datasets on a huge compendium of cancer cell lines to predict synergistic drug combinations. In a case study in an individual breast cancer cell line, MDA-MB-231, we identified drugs selectively active in this cell line compared to the background distribution. We find that target genes of active drugs and other genes involved in related cellular processes were also differentially essential in RNAi screening data. Further, we observe that the drug combinations of these selectively active drugs were synergistic. We propose a systematic approach for integrating chemical screening, drug target and cellular processes knowledge, and functional genetic screens to optimize drug combination testing efforts. Citation Format: Alok Jaiswal, Prson Gautam, Jing Tang, Krister Wennerberg, Tero Aittokallio. Prediction of synergistic anticancer drug combinations by integrating chemical and genetic screens [abstract]. In: Proceedings of the AACR Precision Medicine Series: Opportunities and Challenges of Exploiting Synthetic Lethality in Cancer; Jan 4-7, 2017; San Diego, CA. Philadelphia (PA): AACR; Mol Cancer Ther 2017;16(10 Suppl):Abstract nr A26.

Identifying synergistic drug combinations (SDCs) is a great challenge due to the combinatorial complexity and the fact that SDC is cell line specific. The existing computational methods either did not consider the cell line specificity of SDC, or did not perform well by building model for each cell line independently. In this paper, we present a novel encoder-decoder network named SDCNet for predicting cell line-specific SDCs. SDCNet learns common patterns across different cell lines as well as cell line-specific features in one model for drug combinations. This is realized by considering the SDC graphs of different cell lines as a relational graph, and constructing a relational graph convolutional network (R-GCN) as the encoder to learn and fuse the deep representations of drugs for different cell lines. An attention mechanism is devised to integrate the drug features from different layers of the R-GCN according to their relative importance so that representation learning is further enhanced. The common patterns are exploited through partial parameter sharing in cell line-specific decoders, which not only reconstruct the known SDCs but also predict new ones for each cell line. Experiments on various datasets demonstrate that SDCNet is superior to state-of-the-art methods and is also robust when generalized to new cell lines that are different from the training ones. Finally, the case study again confirms the effectiveness of our method in predicting novel reliable cell line-specific SDCs.

Drug combination therapy is generally more effective than monotherapy in the field of cancer treatment. However, screening for effective synergistic combinations from a wide range of drug combinations is particularly important given the increase in the number of available drug classes and potential drug-drug interactions. Existing methods for predicting the synergistic effects of drug combinations primarily focus on extracting structural features of drug molecules and cell lines, but neglect the interaction mechanisms between cell lines and drug combinations. Consequently, there is a deficiency in comprehensive understanding of the synergistic effects of drug combinations. To address this issue, we propose a drug combination synergy prediction model based on multi-source feature interaction learning, named MFSynDCP, aiming to predict the synergistic effects of anti-tumor drug combinations. This model includes a graph aggregation module with an adaptive attention mechanism for learning drug interactions and a multi-source feature interaction learning controller for managing information transfer between different data sources, accommodating both drug and cell line features. Comparative studies with benchmark datasets demonstrate MFSynDCP's superiority over existing methods. Additionally, its adaptive attention mechanism graph aggregation module identifies drug chemical substructures crucial to the synergy mechanism. Overall, MFSynDCP is a robust tool for predicting synergistic drug combinations. The source code is available from GitHub at https://github.com/kkioplkg/MFSynDCP.

Simple SummaryThis paper introduces a novel graph neural network (a hierarchical graph pooling model), SANEpool, to effectively detect core sub-networks of significant genes for predicting the synergy score of drug/drug combinations in cancer. SANEpool successfully addresses the limitations of the un-transparency in the prediction process of previous computational AI models for drug synergy prediction, while providing the superior predictive performance than popular baselines on numerous drug-synergy prediction datasets. The success of SANEpool indicates that significant gene-gene interactions and gene-drug interactions play a crucial role in designing powerful deep learning models to provide accurate prediction and to reveal the mechanism of the synergy (MoS).Synergistic drug combinations provide huge potentials to enhance therapeutic efficacy and to reduce adverse reactions. However, effective and synergistic drug combination prediction remains an open question because of the unknown causal disease signaling pathways. Though various deep learning (AI) models have been proposed to quantitatively predict the synergism of drug combinations, the major limitation of existing deep learning methods is that they are inherently not interpretable, which makes the conclusions of AI models untransparent to human experts, henceforth limiting the robustness of the model conclusion and the implementation ability of these models in real-world human–AI healthcare. In this paper, we develop an interpretable graph neural network (GNN) that reveals the underlying essential therapeutic targets and the mechanism of the synergy (MoS) by mining the sub-molecular network of great importance. The key point of the interpretable GNN prediction model is a novel graph pooling layer, a self-attention-based node and edge pool (henceforth SANEpool), that can compute the attention score (importance) of genes and connections based on the genomic features and topology. As such, the proposed GNN model provides a systematic way to predict and interpret the drug combination synergism based on the detected crucial sub-molecular network. Experiments on various well-adopted drug-synergy-prediction datasets demonstrate that (1) the SANEpool model has superior predictive ability to generate accurate synergy score prediction, and (2) the sub-molecular networks detected by the SANEpool are self-explainable and salient for identifying synergistic drug combinations.

Transcriptional profiling of drug-treated cells yields high dimensional response signatures that allow drugs to be compared with each other. For example, the Connectivity Map collects signatures that are aggregated across multiple cell types. However, most therapeutic drugs are effective only against a subset of disease genotypes, particularly in the case of anti-cancer drugs. Here we ask how transcriptional signatures vary across cell lines and dose and correlate these signatures to the phenotypic response (growth inhibition). Using these cell line specific signatures, we inferred which signaling pathways are perturbed by specific kinase inhibitors and identified synergistic drug combinations. We treated 6 breast cancer cell lines with more than 100 targeted inhibitors at 6 doses and measured their transcriptional response at 2 time points. We focused on inhibitors targeting key the PI3K and MAPK signaling pathways, as well as receptor tyrosine kinases (RTKs) and cyclin-dependent kinases (CDKs); many of them are currently studied in clinical trials. We identified that ∼40% of the perturbations induce a significant difference in their gene expression profile. Clustering revealed the signatures are time point specific. Some clusters contain perturbations from multiple cell lines, like CDK inhibitors that down regulate genes related to the cell cycle in all six lines. In contrast, clusters comprising inhibitors of the PI3K and MAPK pathways are specific to each cell line and pathway. The perturbations induced by RTK and non-RTK inhibitors cluster with either the PI3K or the MAPK inhibitors depending on the cell line. Thus, the transcriptional response allows us to identify differences in pathway usage between cell lines, in particular to which pathway RTKs signal predominantly. We found that the significance of the transcriptional signature is not necessarily related to growth inhibition. In particular, some inhibitors have little effect on growth, yet induce a significant transcriptional signature. The most striking case is the inhibition of MEK and EGFR in BT20 that induces strong transcriptional and biochemical responses but inhibits growth by only ∼20%. Based on the transcriptional signature we inferred and validated experimentally that FoxO, which is generally regulated by the PI3K pathway, is partially activated by MEK or EGFR inhibition. This suggests that EGFR and PI3K inhibitors act synergistically in BT20, which we validated experimentally both at the level of FoxO activation and growth inhibition. We validated the most promising drug pair by treating xenografts. We have shown how we can use measurements of expression signatures and cellular phenotypes following single drug perturbations to identify drug combinations that are synergistic in individual cell lines. This approach is a step toward the rational design of co-drugging strategies with differential effect and larger therapeutic windows. Citation Format: Marc Hafner, Mario Niepel, Qiaonan Duan, Evan Paull, Josh Stuart, Aravind Subramanian, Avi Ma’ayan, Peter K. Sorger. Transcriptional landscape of drug response guides the design of potent and synergistic drug combinations. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 788.

Depending upon the definition used, as many as half of all cancers diagnosed are considered rare cancers, and most rare cancer patients would greatly benefit from additional therapies. Repurposing of drugs for use in oncology represents an attractive strategy for the rapid development of therapies that address the significant unmet medical need that remains for rare cancers. The FDA-approved anti-helminthic drug mebendazole (MBZ) has been reported to affect several biological pathways involved in tumor growth and metastasis. Based on this activity, multiple rare cancer tumor cell lines including neuroblastoma, carcinoid tumors, esophageal adenocarcinoma, gastrointestinal stromal tumors, leiomyosarcoma, pheochromocytoma, and mesothelioma were analyzed for their susceptibility to MBZ treatment alone or in combination with standard chemotherapeutic agents in vitro. Cultured cells exposed to increasing concentrations of MBZ alone showed a significant decrease in cell viability with IC50s ranging from 0.09-2.2 μM. To evaluate efficacy of MBZ in an additional model system, an ex vivo 3D assay using gastric cancer primary cells isolated from two patient-derived xenograft (PDX) models was performed. Cells were seeded in 3D cultures and treated for seven days with MBZ at nine different doses in triplicate. IC50s were determined to be 0.74 and 0.76 μM indicative of significant potency. These results along with the in vitro data generated using cell lines suggest that MBZ may represent an effective anti-cancer agent across multiple rare cancer indications. However, since tumor cells can often evade or develop resistance to single agent therapy, the potential for synergistic combinations with chemotherapeutic agents was also explored in vitro. Cells were plated in triplicate and treated with a range of concentrations of MBZ and relevant chemotherapeutic agents in a complete Latin square. Combination indexes (CI) were determined based on the method described by Chou and Talalay. Several synergistic drug combinations were identified with CI values ranging from 0.008 to 0.428 (CI < 0.9 indicates synergy). Overall, our results indicate that synergistic combinations of MBZ and chemotherapy agents represent a promising strategy for the treatment of rare cancers. Citation Format: Mahta Samizadeh, Debashree Chakrabarti, Yuhan Zhu, Robert Louis Treuting, Johanne Kaplan, William Siders, Jamie D. Barber. Synergistic activity of mebendazole and chemotherapeutic agents for rare cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 299.

In cancer treatment, drug combinations can convert weakly efficacious monotherapies into regimens that produce robust antitumor activity. O'Neil, Benita, Feldman, and colleagues performed a high-throughput oncology drug combination screen of 583 doublet combinations and found that synergy is relatively rare. However, both well-known and novel synergistic drug combinations were identified. One novel combination, a Wee1 inhibitor and an mTOR inhibitor, potently and synergistically inhibits cancer cell growth both in vitro and in vivo and is well tolerated in vivo. This approach identified novel combinations with improved efficacy that would be difficult to reliably predict based on our current understanding of cancer cell biology.Durable responses of BRAFV600E inhibitor vemurafenib have been hampered by drug resistance in human melanoma. Here, Feng and colleagues created compound DETD-35, a derivative of plant sesquiterpene lactone deoxyelephantopin, which significantly inhibits intrinsic and acquired vemurafenib resistance tumor growth in xenograft mice through suppression of MEK/ERK signaling reactivation and promoting ROS-induced apoptosis in melanoma cells. The antitumor effect of DETD-35 combined with vemurafenib is superior to vemurafenib or DETD-35 treatment alone. These data suggest the therapeutic potential of DETD-35 in BRAFV600E mutant or acquired vemurafenib resistance melanoma as a single or adjuvant agent.Antibody–drug conjugates (ADC) offer increased therapeutic efficacy with reduced side effects compared with chemotherapy. Currier, Ackerman and colleagues explore the role of alternative tumor targeting agents formed through fusion of a disulfide-rich engineered integrin-binding peptide, known as a knottin, to an antibody Fc domain. Knottin–Fc–drug conjugates (KFDC), attached to monomethyl-auristatin-F through a PEG linker, were synthesized via a non-natural amino acid incorporated during cell-free protein synthesis. The KFDC significantly prolonged tumor regression and survival in a xenograft model compared with a nontargeted conjugate or drug alone, thus providing a basis for further development of KFDCs for targeted drug delivery applications.In the present study, Morrison and colleagues discovered for the first time that the transmembrane protein SLITRK6 is overexpressed in bladder cancer. Moreover, the authors investigated SLITRK6 as a target for antibody–drug conjugate (ADC) treatment of advanced urothelial cancer. AGS15E, a SLITRK6-specific ADC conjugated with Monomethyl Auristatin E turned out to be the most potent among all ADCs generated. When compared with its Momomethyl Auristatin F counterpart, AGS15E showed superior antitumor activity in SLITRK6-highly expressed cancer xenografts as well as others with a broad range of SLITRK6 expression. AGS15E is currently in Phase I clinical trials for advanced urothelial cancer.

Increased understanding of the complexity of biological networks is forcing us to reconsider traditional views of disease and treatment. This is particularly true for cancer, as tumor cells have a tremendous capacity to evade the effects of drugs by mutation and/or pathway adaptation. We have developed a combination high throughput screening (cHTS) platform (robotics and analytics) that systematically and efficiently identifies combination drugs with synergistic activity. Synergistic combination drugs can increase maximal drug effect, increase potency and/or circumvent chemoresistance. In addition, coordinated action at multiple molecular targets can provide unique therapeutic benefit not achievable with the “one-drug, one-target” paradigm, especially problematic for cancer. We describe the application of cHTS to the discovery of novel synergistic drug combinations for the treatment of cancer. Using cHTS we have made the surprising discovery that A2A and β2 adrenergic receptor agonists strongly synergize with glucocorticoids (GCs) to inhibit the proliferation of hematologic malignancies. Using an 83 cell line panel, we find that synergistic activity is specific for select B-cell malignancies (including some cell lines that are GC-insensitive). These dexamethasone-dependent synergies are observed across a large panel of multiple myeloma (MM) cell lines and in multiple complex preclinical models of MM disease. In general, chemoresistance is a recurrent problem for cancer drugs and development of resistance after chronic exposure can reduce drug efficacy and promote refractory disease. We therefore examined the effects of chronic exposure to either A2A or β2AR agonists. Exposure of MM cells (MM.1S) to CGS-21680 (A2A agonist) or salmeterol (β2AR agonist) for one month reduced single agent sensitivity >80%. Surprisingly, combinations of either agent with dexamethasone maintained similar amounts of synergy and cell killing as found with naïve untreated cells. Our studies demonstrate that synergistic combinations of A2A and β2AR agonists are highly selective for B-cell malignancies and support the notion that synergistic drug combinations can improve therapeutically relevant selectivity and circumvent drug resistance. These preclinical studies reinforce the rationale for investigation of A2A and β2AR agonists in the treatment of B-cell malignancies and in particular, patients who have MM. Furthermore, this work highlights the power of cHTS to interrogate combination activity across large panels of cancer cell lines with distinct molecular phenotypes and to identify novel mechanisms for therapeutic application. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 5509.

Oral cancer is one of the most common forms of cancer among developing countries but despite current efforts, the survival of oral cancer has remained constant. Current research on oral cancer is centered on clinical studies focused on single drugs or very few combinations of the most common therapeutics historically used to treat oral cancer. However, there is a major gap in the literature about combinations of drug treatments for oral cancer. The goal of this study is to explore combination treatments and assess whether any synergistic effects can be produced. To accomplish this, two oral squamous cell carcinoma (OSCC) cell lines, Cal‐27 and OECM‐1 from the tongue and the oral cavity, respectively, will be used to investigate the efficacy of single drugs as well as combinations of drugs from a wide array of chemotherapy agents currently used to treat oral cancer as well as other forms of cancer. Thus far, Cell Titer Glo assays have been performed in triplicate for both Cal‐27 and OECM‐1 to analyze their viability in the presence of four drugs used to treat other forms of cancer with different mechanisms of action: PD0, a MEK inhibitor, IBET‐762, a BET inhibitor, ABT‐199, a Bcl‐2 inhibitor, and JQ1, a BRD4 inhibitor. GI50 values were gathered from each assay. Cal‐27 presented higher GI50 values than OECM‐1 for IBET‐762 and JQ1 at 1.52 mM and 1.64 mM, respectively. The GI50 values for OECM‐1 treatment with IBET‐762 and JQ1 were lower at 1.07 mM and 1.21 mM, respectively. Both cell lines were found to have similar resistance to PD0, around 1.05 mM, and ABT‐199, around 1.02 mM. High GI50 values across the cell lines indicate that both Cal‐27 and OECM‐1 are resistant to the four drugs that were tested. In the future, Cal‐27 and OECM‐1 will be treated with a panel of 10 drugs currently used clinically to treat oral cancer to gather baseline data to understand how these cell lines react to current treatment. The drugs that will be used include: Cisplatin, 5‐FU, Taxol, Docetaxel, Hydroxyurea, Methotrexate, Bleomycin, Gemcitabine, and Vincristine, all of which interfere with DNA synthesis and cell growth. Then, Cal‐27 and OECM‐1 will be treated with new combinations of drugs currently used to treat oral cancer and drugs not currently used for oral cancer but used to treat other forms of cancer. Some of these drugs include: Gefitinib, a EGFR inhibitor, Venetoclax, a BCL‐2 inhibitor, Irinotecan, a topoisomerase inhibitor, Mitoxantrone, a topoisomerase 2 and PKC inhibitor, and several other drugs. Cellular responses will then be assessed to determine if any combinations are synergistic. The data generated can aid researchers in understanding how different OSCC cells respond to drug combinations, which may lead to more successful clinical treatment options in the future.

Intrinsic and acquired drug resistance is a major challenge in cancer therapy. Synergistic drug combinations could help to overcome drug resistance. However, the number of possible drug combinations is enormous, and it is infeasible to experimentally screen all drug combinations with limited resources. Therefore, computational models to predict and prioritize effective drug combinations are important for combination therapy discovery. Compared with existing models, we propose a novel deep learning model, AuDNNsynergy, to predict the synergy of pairwise drug combinations by integrating multiomics data. Specifically, three autoencoders are trained using the gene expression, copy number, and genetic mutation data of tumor samples from The Cancer Genome Atlas (TCGA). Then the gene expression, copy number, and mutation of individual cancer cell lines are coded using the three trained autoencoders. The physicochemical features of individual drugs and the encoded omics data of individual cancer cell lines are used as the input features of a deep neural network that predicts the synergy score of given pairwise drug combinations against the specific cancer cell lines. The comparison results showed the proposed AuDNNsynergy model outperforms, specifically in terms of rank correlation metric, four state-of-the-art approaches, namely, DeepSynergy, Gradient Boosting Machines, Random Forests, and Elastic Nets.