Landscape and function of multiple mutations within individual oncogenes.

Sporadic reports have described cancer cases in which multiple driver mutations (MMs) occur in the same oncogene1,2. However, the overall landscape and relevance of MMs remain elusive. Here we carried out a pan-cancer analysis of 60,954 cancer samples, and identified 14 pan-cancer and 6 cancer-type-specific oncogenes in which MMs occur more frequently than expected: 9% of samples with at least one mutation in these genes harboured MMs. In various oncogenes, MMs are preferentially present in cis and show markedly different mutational patterns compared with single mutations in terms of type (missense mutations versus in-frame indels), position and amino-acid substitution, suggesting a cis-acting effect on mutational selection. MMs show an overrepresentation of functionally weak, infrequent mutations, which confer enhanced oncogenicity in combination. Cells with MMs in the PIK3CA and NOTCH1 genes exhibit stronger dependencies on the mutated genes themselves, enhanced downstream signalling activation and/or greater sensitivity to inhibitory drugs than those with single mutations. Together oncogenic MMs are a relatively common driver event, providing the underlying mechanism for clonal selection of suboptimal mutations that are individually rare but collectively account for a substantial proportion of oncogenic mutations.

Artificial Intelligence-Based Histopathological Subtyping of High-Grade Serous Ovarian Cancer.

Four subtypes of ovarian high-grade serous carcinoma (HGSC) have previously been identified, each with different prognoses and drug sensitivities. However, the accuracy of classification depended on the assessor's experience. This study aimed to develop a universal algorithm for HGSC-subtype classification using deep learning techniques. An artificial intelligence (AI)-based classification algorithm, which replicates the consensus diagnosis of pathologists, was formulated to analyze the morphological patterns and tumor-infiltrating lymphocyte counts for each tile extracted from whole slide images of ovarian HGSC available in The Cancer Genome Atlas (TCGA) data set. The accuracy of the algorithm was determined using the validation set from the Japanese Gynecologic Oncology Group 3022A1 (JGOG3022A1) and Kindai and Kyoto University (Kindai/Kyoto) cohorts. The algorithm classified the four HGSC-subtypes with mean accuracies of 0.933, 0.910, and 0.862 for the TCGA, JGOG3022A1, and Kindai/Kyoto cohorts, respectively. To compare mesenchymal transition (MT) with non-MT groups, overall survival analysis was performed in the TCGA data set. The AI-based prediction of HGSC-subtype classification in TCGA cases showed that the MT group had a worse prognosis than the non-MT group (P = 0.017). Furthermore, Cox proportional hazard regression analysis identified AI-based MT subtype classification prediction as a contributing factor along with residual disease after surgery, stage, and age. In conclusion, a robust AI-based HGSC-subtype classification algorithm was established using virtual slides of ovarian HGSC.

Decoding the specificity of m6A RNA methylation and its implication in cancer therapy.

N6-methyladenosine (m6A) is the most abundant endogenous modification in eukaryotic RNAs. It plays important roles in various biological processes and diseases, including cancers. More and more studies have revealed that the deposition of m6A is specifically regulated in a context-dependent manner. Here, we review the diverse mechanisms that determine the topology of m6A along RNAs and the cell-type-specific m6A methylomes. The exon junction complex (EJC) as well as histone modifications play important roles in determining the topological distribution of m6A along nascent RNAs, while the transcription factors and RNA-binding proteins, which usually bind specific DNAs and RNAs in a cell-type-specific manner, largely account for the cell-type-specific m6A methylomes. Due to the lack of specificity of m6A writers and readers, there are still challenges to target the core m6A machinery for cancer therapies. Therefore, understanding the mechanisms underlying the specificity of m6A modifications in cancers would be important for future cancer therapies through m6A intervention.

Molecular Dynamics Unveils Multiple-Site Binding of Inhibitors with Reduced Activity on the Surface of Dihydrofolate Reductase.

The sensitivity to protein inhibitors is altered by modifications or protein mutations, as represented by drug resistance. The mode of stable drug binding to the protein pocket has been experimentally clarified. However, the nature of the binding of inhibitors with reduced sensitivity remains unclear at the atomic level. In this study, we analyzed the thermodynamics and kinetics of inhibitor binding to the surface of wild-type and mutant dihydrofolate reductase (DHFR) using molecular dynamics simulations combined with Markov state modeling. A strong inhibitor of methotrexate (MTX) showed a preference for the active site of wild-type DHFR with minimal binding to unrelated (secondary) sites. Deletion of a side-chain fragment in MTX largely destabilized the active site-bound state, with clear evidence of binding to secondary sites. Similarly, the F31V mutation in DHFR diminished the specificity of MTX binding to the active site. These results reveal the presence of multiple-bound states whose stabilities are comparable to or higher than those of the unbound state, suggesting that a reduction in the binding affinity for the active site significantly elevates the fractions of these states. This study presents a theoretical model that more accurately interprets the altered drug sensitivity than the traditional two-state model.

Binding Mechanism between Platelet Glycoprotein and Cyclic Peptide Elucidated by McMD-Based Dynamic Docking.

The cyclic peptide OS1 (amino acid sequence: CTERMALHNLC), which has a disulfide bond between both termini cysteine residues, inhibits complex formation between the platelet glycoprotein Ibα (GPIbα) and the von Willebrand factor (vWF) by forming a complex with GPIbα. To study the binding mechanism between GPIbα and OS1 and, therefore, the inhibition mechanism of the protein-protein GPIbα-vWF complex, we have applied our multicanonical molecular dynamics (McMD)-based dynamic docking protocol starting from the unbound state of the peptide. Our simulations have reproduced the experimental complex structure, although the top-ranking structure was an intermediary one, where the peptide was bound in the same location as in the experimental structure; however, the ß-switch of GPIbα attained a different conformation. Our analysis showed that subsequent refolding of the ß-switch results in a more stable binding configuration, although the transition to the native configuration appears to take some time, during which OS1 could dissociate. Our results show that conformational changes in the ß-switch are crucial for successful binding of OS1. Furthermore, we identified several allosteric binding sites of GPIbα that might also interfere with vWF binding, and optimization of the peptide to target these allosteric sites might lead to a more effective inhibitor, as these are not dependent on the ß-switch conformation.

Measurement of the Acetabular Cup Orientation After Total Hip Arthroplasty Based on 3-Dimensional Reconstruction From a Single X-Ray Image Using Generative Adversarial Networks.

BACKGROUND: The purpose of this study was to reconstruct 3-dimensional (3D) computed tomography (CT) images from single anteroposterior (AP) postoperative total hip arthroplasty (THA) X-ray images using a deep learning algorithm known as generative adversarial networks (GANs) and to validate the accuracy of cup angle measurement on GAN-generated CT. METHODS: We used 2 GAN-based models, CycleGAN and X2CT-GAN, to generate 3D CT images from X-ray images of 386 patients who underwent primary THAs using a cementless cup. The training dataset consisted of 522 CT images and 2,282 X-ray images. The image quality was validated using the peak signal-to-noise ratio and the structural similarity index measure. The cup anteversion and inclination measurements on the GAN-generated CT images were compared with the actual CT measurements. Statistical analyses of absolute measurement errors were performed using Mann-Whitney U tests and nonlinear regression analyses. RESULTS: The study successfully achieved 3D reconstruction from single AP postoperative THA X-ray images using GANs, exhibiting excellent peak signal-to-noise ratio (37.40) and structural similarity index measure (0.74). The median absolute difference in radiographic anteversion was 3.45° and the median absolute difference in radiographic inclination was 3.25°, respectively. Absolute measurement errors tended to be larger in cases with cup malposition than in those with optimal cup orientation. CONCLUSIONS: This study demonstrates the potential of GANs for 3D reconstruction from single AP postoperative THA X-ray images to evaluate cup orientation. Further investigation and refinement of this model are required to improve its performance.

Network-based prediction approach for cancer-specific driver missense mutations using a graph neural network.

BACKGROUND: In cancer genomic medicine, finding driver mutations involved in cancer development and tumor growth is crucial. Machine-learning methods to predict driver missense mutations have been developed because variants are frequently detected by genomic sequencing. However, even though the abnormalities in molecular networks are associated with cancer, many of these methods focus on individual variants and do not consider molecular networks. Here we propose a new network-based method, Net-DMPred, to predict driver missense mutations considering molecular networks. Net-DMPred consists of the graph part and the prediction part. In the graph part, molecular networks are learned by a graph neural network (GNN). The prediction part learns whether variants are driver variants using features of individual variants combined with the graph features learned in the graph part. RESULTS: Net-DMPred, which considers molecular networks, performed better than conventional methods. Furthermore, the prediction performance differed by the molecular network structure used in learning, suggesting that it is important to consider not only the local network related to cancer but also the large-scale network in living organisms. CONCLUSIONS: We propose a network-based machine learning method, Net-DMPred, for predicting cancer driver missense mutations. Our method enables us to consider the entire graph architecture representing the molecular network because it uses GNN. Net-DMPred is expected to detect driver mutations from a lot of missense mutations that are not known to be associated with cancer.

Inhibition of the bone morphogenetic protein pathway suppresses tumor growth through downregulation of epidermal growth factor receptor in MEK/ERK-dependent colorectal cancer.

The bone morphogenetic protein (BMP) pathway promotes differentiation and induces apoptosis in normal colorectal epithelial cells. However, its role in colorectal cancer (CRC) is controversial, where it can act as context-dependent tumor promoter or tumor suppressor. Here we have found that CRC cells reside in a BMP-rich environment based on curation of two publicly available RNA-sequencing databases. Suppression of BMP using a specific BMP inhibitor, LDN193189, suppresses the growth of select CRC organoids. Colorectal cancer organoids treated with LDN193189 showed a decrease in epidermal growth factor receptor, which was mediated by protein degradation induced by leucine-rich repeats and immunoglobulin-like domains protein 1 (LRIG1) expression. Among 18 molecularly characterized CRC organoids, suppression of growth by BMP inhibition correlated with induction of LRIG1 gene expression. Notably, knockdown of LRIG1 in organoids diminished the growth-suppressive effect of LDN193189. Furthermore, in CRC organoids, which are susceptible to growth suppression by LDN193189, simultaneous treatment with LDN193189 and trametinib, an FDA-approved MEK inhibitor, resulted in cooperative growth inhibition both in vitro and in vivo. Taken together, the simultaneous inhibition of BMP and MEK could be a novel treatment option in CRC cases, and evaluating in vitro growth suppression and LRIG1 induction by BMP inhibition using patient-derived organoids could offer functional biomarkers for predicting potential responders to this regimen.

Deep Learning-Based Differential Diagnosis of Follicular Thyroid Tumors Using Histopathological Images.

Deep learning systems (DLSs) have been developed for the histopathological assessment of various types of tumors, but none are suitable for differential diagnosis between follicular thyroid carcinoma (FTC) and follicular adenoma (FA). Furthermore, whether DLSs can identify the malignant characteristics of thyroid tumors based only on random views of tumor tissue histology has not been evaluated. In this study, we developed DLSs able to differentiate between FTC and FA based on 3 types of convolutional neural network architecture: EfficientNet, VGG16, and ResNet50. The performance of all 3 DLSs was excellent (area under the receiver operating characteristic curve = 0.91 ± 0.04; F1 score = 0.82 ± 0.06). Visual explanations using gradient-weighted class activation mapping suggested that the diagnosis of both FTC and FA was largely dependent on nuclear features. The DLSs were then trained with FTC images and linked information (presence or absence of recurrence within 10 years, vascular invasion, and wide capsular invasion). The ability of the DLSs to diagnose these characteristics was then determined. The results showed that, based on the random views of histology, the DLSs could predict the risk of FTC recurrence, vascular invasion, and wide capsular invasion with a certain level of accuracy (area under the receiver operating characteristic curve = 0.67 ± 0.13, 0.62 ± 0.11, and 0.65 ± 0.09, respectively). Further improvement of our DLSs could lead to the establishment of automated differential diagnosis systems requiring only biopsy specimens.

CBNplot: Bayesian network plots for enrichment analysis.

SUMMARY: When investigating gene expression profiles, determining important directed edges between genes can provide valuable insights in addition to identifying differentially expressed genes. In the subsequent functional enrichment analysis (EA), understanding how enriched pathways or genes in the pathway interact with one another can help infer the gene regulatory network (GRN), important for studying the underlying molecular mechanisms. However, packages for easy inference of the GRN based on EA are scarce. Here, we developed an R package, CBNplot, which infers the Bayesian network (BN) from gene expression data, explicitly utilizing EA results obtained from curated biological pathway databases. The core features include convenient wrapping for structure learning, visualization of the BN from EA results, comparison with reference networks, and reflection of gene-related information on the plot. As an example, we demonstrate the analysis of bladder cancer-related datasets using CBNplot, including probabilistic reasoning, which is a unique aspect of BN analysis. We display the transformability of results obtained from one dataset to another, the validity of the analysis as assessed using established knowledge and literature, and the possibility of facilitating knowledge discovery from gene expression datasets. AVAILABILITY AND IMPLEMENTATION: The library, documentation and web server are available at https://github.com/noriakis/CBNplot. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Improving Compound-Protein Interaction Prediction by Self-Training with Augmenting Negative Samples.

Identifying compound-protein interactions (CPIs) is crucial for drug discovery. Since experimentally validating CPIs is often time-consuming and costly, computational approaches are expected to facilitate the process. Rapid growths of available CPI databases have accelerated the development of many machine-learning methods for CPI predictions. However, their performance, particularly their generalizability against external data, often suffers from a data imbalance attributed to the lack of experimentally validated inactive (negative) samples. In this study, we developed a self-training method for augmenting both credible and informative negative samples to improve the performance of models impaired by data imbalances. The constructed model demonstrated higher performance than those constructed with other conventional methods for solving data imbalances, and the improvement was prominent for external datasets. Moreover, examination of the prediction score thresholds for pseudo-labeling during self-training revealed that augmenting the samples with ambiguous prediction scores is beneficial for constructing a model with high generalizability. The present study provides guidelines for improving CPI predictions on real-world data, thus facilitating drug discovery.

VGAE-MCTS: A New Molecular Generative Model Combining the Variational Graph Auto-Encoder and Monte Carlo Tree Search.

Molecular generation is crucial for advancing drug discovery, materials science, and chemical exploration. It expedites the search for new drug candidates, facilitates tailored material creation, and enhances our understanding of molecular diversity. By employing artificial intelligence techniques such as molecular generative models based on molecular graphs, researchers have tackled the challenge of identifying efficient molecules with desired properties. Here, we propose a new molecular generative model combining a graph-based deep neural network and a reinforcement learning technique. We evaluated the validity, novelty, and optimized physicochemical properties of the generated molecules. Importantly, the model explored uncharted regions of chemical space, allowing for the efficient discovery and design of new molecules. This innovative approach has considerable potential to revolutionize drug discovery, materials science, and chemical research for accelerating scientific innovation. By leveraging advanced techniques and exploring previously unexplored chemical spaces, this study offers promising prospects for the efficient discovery and design of new molecules in the field of drug development.

Individual health-disease phase diagrams for disease prevention based on machine learning.

Early disease detection and prevention methods based on effective interventions are gaining attention worldwide. Progress in precision medicine has revealed that substantial heterogeneity exists in health data at the individual level and that complex health factors are involved in chronic disease development. Machine-learning techniques have enabled precise personal-level disease prediction by capturing individual differences in multivariate data. However, it is challenging to identify what aspects should be improved for disease prevention based on future disease-onset prediction because of the complex relationships among multiple biomarkers. Here, we present a health-disease phase diagram (HDPD) that represents an individual's health state by visualizing the future-onset boundary values of multiple biomarkers that fluctuate early in the disease progression process. In HDPDs, future-onset predictions are represented by perturbing multiple biomarker values while accounting for dependencies among variables. We constructed HDPDs for 11 diseases using longitudinal health checkup cohort data of 3,238 individuals, comprising 3,215 measurement items and genetic data. The improvement of biomarker values to the non-onset region in HDPD remarkably prevented future disease onset in 7 out of 11 diseases. HDPDs can represent individual physiological states in the onset process and be used as intervention goals for disease prevention.

eSkip-Finder: a machine learning-based web application and database to identify the optimal sequences of antisense oligonucleotides for exon skipping.

Exon skipping using antisense oligonucleotides (ASOs) has recently proven to be a powerful tool for mRNA splicing modulation. Several exon-skipping ASOs have been approved to treat genetic diseases worldwide. However, a significant challenge is the difficulty in selecting an optimal sequence for exon skipping. The efficacy of ASOs is often unpredictable, because of the numerous factors involved in exon skipping. To address this gap, we have developed a computational method using machine-learning algorithms that factors in many parameters as well as experimental data to design highly effective ASOs for exon skipping. eSkip-Finder (https://eskip-finder.org) is the first web-based resource for helping researchers identify effective exon skipping ASOs. eSkip-Finder features two sections: (i) a predictor of the exon skipping efficacy of novel ASOs and (ii) a database of exon skipping ASOs. The predictor facilitates rapid analysis of a given set of exon/intron sequences and ASO lengths to identify effective ASOs for exon skipping based on a machine learning model trained by experimental data. We confirmed that predictions correlated well with in vitro skipping efficacy of sequences that were not included in the training data. The database enables users to search for ASOs using queries such as gene name, species, and exon number.

Development and validation of ischemic heart disease and stroke prognostic models using large-scale real-world data from Japan.

BACKGROUND: Previous cardiovascular risk prediction models in Japan have utilized prospective cohort studies with concise data. As the health information including health check-up records and administrative claims becomes digitalized and publicly available, application of large datasets based on such real-world data can achieve prediction accuracy and support social implementation of cardiovascular disease risk prediction models in preventive and clinical practice. In this study, classical regression and machine learning methods were explored to develop ischemic heart disease (IHD) and stroke prognostic models using real-world data. METHODS: IQVIA Japan Claims Database was searched to include 691,160 individuals (predominantly corporate employees and their families working in secondary and tertiary industries) with at least one annual health check-up record during the identification period (April 2013-December 2018). The primary outcome of the study was the first recorded IHD or stroke event. Predictors were annual health check-up records at the index year-month, comprising demographic characteristics, laboratory tests, and questionnaire features. Four prediction models (Cox, Elnet-Cox, XGBoost, and Ensemble) were assessed in the present study to develop a cardiovascular disease risk prediction model for Japan. RESULTS: The analysis cohort consisted of 572,971 invididuals. All prediction models showed similarly good performance. The Harrell's C-index was close to 0.9 for all IHD models, and above 0.7 for stroke models. In IHD models, age, sex, high-density lipoprotein, low-density lipoprotein, cholesterol, and systolic blood pressure had higher importance, while in stroke models systolic blood pressure and age had higher importance. CONCLUSION: Our study analyzed classical regression and machine learning algorithms to develop cardiovascular disease risk prediction models for IHD and stroke in Japan that can be applied to practical use in a large population with predictive accuracy.

Protein-ligand binding affinity prediction of cyclin-dependent kinase-2 inhibitors by dynamically averaged fragment molecular orbital-based interaction energy.

Fragment molecular orbital (FMO) method is a powerful computational tool for structure-based drug design, in which protein-ligand interactions can be described by the inter-fragment interaction energy (IFIE) and its pair interaction energy decomposition analysis (PIEDA). Here, we introduced a dynamically averaged (DA) FMO-based approach in which molecular dynamics simulations were used to generate multiple protein-ligand complex structures for FMO calculations. To assess this approach, we examined the correlation between the experimental binding free energies and DA-IFIEs of six CDK2 inhibitors whose net charges are zero. The correlation between the experimental binding free energies and snapshot IFIEs for X-ray crystal structures was R2  = 0.75. Using the DA-IFIEs, the correlation significantly improved to 0.99. When an additional CDK2 inhibitor with net charge of -1 was added, the DA FMO-based scheme with the dispersion energies still achieved R2  = 0.99, whereas R2 decreased to 0.32 employing all the energy terms of PIEDA.

Single-Image Super-Resolution Improvement of X-ray Single-Particle Diffraction Images Using a Convolutional Neural Network.

Femtosecond X-ray pulse lasers are promising probes for the elucidation of the multiconformational states of biomolecules because they enable snapshots of single biomolecules to be observed as coherent diffraction images. Multi-image processing using an X-ray free-electron laser has proven to be a successful structural analysis method for viruses. However, the performance of single-particle analysis (SPA) for flexible biomolecules with sizes ≤100 nm remains difficult. Owing to the multiconformational states of biomolecules and noisy character of diffraction images, diffraction image improvement by multi-image processing is often ineffective for such molecules. Herein, a single-image super-resolution (SR) model was constructed using an SR convolutional neural network (SRCNN). Data preparation was performed in silico to consider the actual observation situation with unknown molecular orientations and the fluctuation of molecular structure and incident X-ray intensity. It was demonstrated that the trained SRCNN model improved the single-particle diffraction image quality, corresponding to an observed image with an incident X-ray intensity (approximately three to seven times higher than the original X-ray intensity), while retaining the individuality of the diffraction images. The feasibility of SPA for flexible biomolecules with sizes ≤100 nm was dramatically increased by introducing the SRCNN improvement at the beginning of the various structural analysis schemes.

AI-Driven Synthetic Route Design Incorporated with Retrosynthesis Knowledge.

Computer-aided synthesis planning (CASP) aims to assist chemists in performing retrosynthetic analysis for which they utilize their experiments, intuition, and knowledge. Recent breakthroughs in machine learning (ML) techniques, including deep neural networks, have significantly improved data-driven synthetic route designs without human intervention. However, learning chemical knowledge by ML for practical synthesis planning has not yet been adequately achieved and remains a challenging problem. In this study, we developed a data-driven CASP application integrated with various portions of retrosynthesis knowledge called "ReTReK" that introduces the knowledge as adjustable parameters into the evaluation of promising search directions. The experimental results showed that ReTReK successfully searched synthetic routes based on the specified retrosynthesis knowledge, indicating that the synthetic routes searched with the knowledge were preferred to those without the knowledge. The concept of integrating retrosynthesis knowledge as adjustable parameters into a data-driven CASP application is expected to enhance the performance of both existing data-driven CASP applications and those under development.

Repertoires of G protein-coupled receptors for Ciona-specific neuropeptides.

Neuropeptides play pivotal roles in various biological events in the nervous, neuroendocrine, and endocrine systems, and are correlated with both physiological functions and unique behavioral traits of animals. Elucidation of functional interaction between neuropeptides and receptors is a crucial step for the verification of their biological roles and evolutionary processes. However, most receptors for novel peptides remain to be identified. Here, we show the identification of multiple G protein-coupled receptors (GPCRs) for species-specific neuropeptides of the vertebrate sister group, Ciona intestinalis Type A, by combining machine learning and experimental validation. We developed an original peptide descriptor-incorporated support vector machine and used it to predict 22 neuropeptide-GPCR pairs. Of note, signaling assays of the predicted pairs identified 1 homologous and 11 Ciona-specific neuropeptide-GPCR pairs for a 41% hit rate: the respective GPCRs for Ci-GALP, Ci-NTLP-2, Ci-LF-1, Ci-LF-2, Ci-LF-5, Ci-LF-6, Ci-LF-7, Ci-LF-8, Ci-YFV-1, and Ci-YFV-3. Interestingly, molecular phylogenetic tree analysis revealed that these receptors, excluding the Ci-GALP receptor, were evolutionarily unrelated to any other known peptide GPCRs, confirming that these GPCRs constitute unprecedented neuropeptide receptor clusters. Altogether, these results verified the neuropeptide-GPCR pairs in the protochordate and evolutionary lineages of neuropeptide GPCRs, and pave the way for investigating the endogenous roles of novel neuropeptides in the closest relatives of vertebrates and the evolutionary processes of neuropeptidergic systems throughout chordates. In addition, the present study also indicates the versatility of the machine-learning-assisted strategy for the identification of novel peptide-receptor pairs in various organisms.

Molecular dynamics simulation-guided drug sensitivity prediction for lung cancer with rare EGFR mutations.

Next generation sequencing (NGS)-based tumor profiling identified an overwhelming number of uncharacterized somatic mutations, also known as variants of unknown significance (VUS). The therapeutic significance of EGFR mutations outside mutational hotspots, consisting of >50 types, in nonsmall cell lung carcinoma (NSCLC) is largely unknown. In fact, our pan-nation screening of NSCLC without hotspot EGFR mutations (n = 3,779) revealed that the majority (>90%) of cases with rare EGFR mutations, accounting for 5.5% of the cohort subjects, did not receive EGFR-tyrosine kinase inhibitors (TKIs) as a first-line treatment. To tackle this problem, we applied a molecular dynamics simulation-based model to predict the sensitivity of rare EGFR mutants to EGFR-TKIs. The model successfully predicted the diverse in vitro and in vivo sensitivities of exon 20 insertion mutants, including a singleton, to osimertinib, a third-generation EGFR-TKI (R2 = 0.72, P = 0.0037). Additionally, our model showed a higher consistency with experimentally obtained sensitivity data than other prediction approaches, indicating its robustness in analyzing complex cancer mutations. Thus, the in silico prediction model will be a powerful tool in precision medicine for NSCLC patients carrying rare EGFR mutations in the clinical setting. Here, we propose an insight to overcome mutation diversity in lung cancer.