MaltSci 麦伴科研

Appearance

Popular Reports / computational-biology

This report is written by MaltSci based on the latest literature and research findings

What are the applications of deep learning in biology? ​

Abstract ​

Deep learning has emerged as a transformative force in the field of biology, offering significant advancements in various applications such as genomics, proteomics, drug discovery, and personalized medicine. The ability of deep learning algorithms to analyze and interpret vast amounts of complex biological data has enabled researchers to uncover intricate patterns and relationships that were previously challenging to identify using traditional methods. In genomics, deep learning techniques are utilized for genome annotation, variant calling, and gene expression analysis, enhancing our understanding of genetic information and facilitating precision medicine approaches. In proteomics, deep learning aids in protein structure prediction and the analysis of protein-protein interactions, contributing to advancements in drug discovery and therapeutic design. Furthermore, deep learning has transformed the drug discovery process by improving drug target identification and predictive modeling of drug responses, streamlining the development of new therapeutics. Despite these advancements, challenges related to data quality, model interpretability, and ethical implications persist, necessitating ongoing research and development. The future of deep learning in biology promises further integration with other computational methods and the exploration of multi-omics data, which will enhance diagnostic precision and treatment personalization. This report synthesizes current literature on deep learning applications in biology, highlighting the methodology, impact, challenges, and future directions for this rapidly evolving field.

Outline ​

This report will discuss the following questions.

  • 1 Introduction
  • 2 Overview of Deep Learning Techniques
    • 2.1 Neural Networks and Their Variants
    • 2.2 Training and Optimization Strategies
  • 3 Applications in Genomics
    • 3.1 Genome Annotation and Variant Calling
    • 3.2 Gene Expression Analysis
  • 4 Applications in Proteomics
    • 4.1 Protein Structure Prediction
    • 4.2 Protein-Protein Interaction Prediction
  • 5 Applications in Drug Discovery
    • 5.1 Drug Target Identification
    • 5.2 Predictive Modeling of Drug Responses
  • 6 Challenges and Limitations
    • 6.1 Data Quality and Availability
    • 6.2 Interpretability of Deep Learning Models
  • 7 Future Directions
    • 7.1 Integration with Other Computational Methods
    • 7.2 Ethical Considerations in Biological Research
  • 8 Summary

1 Introduction ​

Deep learning, a powerful subset of artificial intelligence, has emerged as a transformative force across numerous disciplines, including biology. The capacity of deep learning algorithms to process and analyze vast datasets has enabled significant advancements in various biological fields, such as genomics, proteomics, drug discovery, and personalized medicine. The ability to identify intricate patterns within high-dimensional data makes deep learning particularly suited for tackling complex biological questions that traditional computational methods struggle to address. As such, understanding the applications and implications of deep learning in biology is crucial for both researchers and practitioners aiming to harness its potential to revolutionize biological research and healthcare.

The significance of deep learning in biology cannot be overstated. As biological research generates increasingly large and complex datasets—from genomic sequences to cellular images—the need for sophisticated analytical tools has become paramount. Deep learning has demonstrated its ability to improve predictive accuracy and efficiency in various biological applications, thereby facilitating more informed decision-making in clinical settings and advancing our understanding of fundamental biological processes [1][2]. The integration of deep learning into biological research represents not only a methodological shift but also a paradigm change that has the potential to redefine the landscape of biomedical research and healthcare delivery.

Currently, the application of deep learning in biology is witnessing rapid growth, characterized by a diverse range of methodologies and applications. In genomics, deep learning techniques are employed for tasks such as genome annotation, variant calling, and gene expression analysis, showcasing the technology's ability to decode the complexities of genetic information [3][4]. In proteomics, deep learning is utilized for protein structure prediction and the analysis of protein-protein interactions, further highlighting its versatility in addressing key biological questions [5]. The drug discovery process has also been transformed through deep learning, which aids in drug target identification and the predictive modeling of drug responses, thus streamlining the development of new therapeutics [6].

Despite the promising advancements, the integration of deep learning into biological research is not without challenges. Issues related to data quality and availability, the interpretability of deep learning models, and the ethical implications of using these technologies in research and clinical practice remain critical concerns [1][7]. Addressing these challenges is essential for the continued advancement of deep learning applications in biology and for ensuring that these technologies are applied responsibly and effectively.

This report is organized as follows: the next section provides an overview of deep learning techniques, including neural networks and their variants, along with training and optimization strategies. Subsequent sections delve into the specific applications of deep learning in genomics, proteomics, and drug discovery, highlighting key methodologies and their impacts. We will then discuss the challenges and limitations faced in this field, followed by a look at future directions for research and development. Finally, the report concludes with a summary of the findings and their implications for the future of deep learning in biology. By synthesizing the current literature and offering insights into the state of deep learning applications, this report aims to enhance understanding and foster further exploration of this rapidly evolving field.

2 Overview of Deep Learning Techniques ​

2.1 Neural Networks and Their Variants ​

Deep learning has emerged as a transformative approach in the field of biology, leveraging its capabilities to analyze complex and high-dimensional data. This technology is particularly effective in extracting meaningful patterns from large datasets, which are common in biological research. Below is an overview of the applications of deep learning in various biological contexts, along with a discussion of neural networks and their variants.

Deep learning techniques, particularly those derived from artificial neural networks, have shown significant promise in several areas of biology. One notable application is in the analysis of genomic sequences, where deep learning models have been employed to predict the structure and function of genomic elements, such as promoters and enhancers, as well as gene expression levels [2]. These applications illustrate how deep learning can enhance our understanding of complex biological systems.

In the realm of medical imaging, deep learning has revolutionized the classification and analysis of medical images, facilitating advancements in diagnostics and patient management [8]. For instance, convolutional neural networks (CNNs) are frequently utilized for tasks such as tumor detection and classification in radiology [9]. The ability of these networks to learn hierarchical features from images enables more accurate and efficient diagnoses compared to traditional methods.

Another significant application of deep learning is in cancer research, where it aids in the analysis of omics data, including genomic, methylation, and transcriptomic data. Deep learning models are being developed to integrate these diverse data types to support decision-making in cancer diagnosis and treatment [9]. This integration is crucial for developing personalized medicine approaches, as it allows for a more comprehensive understanding of individual patient profiles.

Deep learning has also been applied to protein structure prediction and antibody design, marking a significant advancement in drug discovery and development [5]. The integration of deep learning with computational methods accelerates the identification of lead candidates and optimizes the design of therapeutic proteins, which is vital in addressing complex diseases.

In the context of plant biology, deep learning techniques have been utilized to analyze plant genome sequences, predict gene expression, and identify epigenetic features [4]. These applications enhance our understanding of plant genetics and can lead to innovations in agriculture, such as the development of genetically modified crops with improved traits.

Neural networks, the backbone of deep learning, consist of layers of interconnected nodes (neurons) that process input data. Variants of neural networks, such as recurrent neural networks (RNNs) and CNNs, are tailored for specific types of data. RNNs are particularly suited for sequential data, such as time-series data or genomic sequences, while CNNs excel in processing grid-like data, such as images [10].

The architecture of neural networks can be optimized through various techniques, including dropout, batch normalization, and data augmentation, which enhance model performance and generalization [11]. Moreover, advancements in computational power and algorithmic strategies have made it feasible to apply deep learning to increasingly complex biological datasets, opening new avenues for research and discovery [12].

In summary, deep learning is reshaping the landscape of biological research by providing powerful tools for data analysis and interpretation across multiple domains, including genomics, medical imaging, cancer research, and plant biology. The continued evolution of neural network architectures and their applications promises to drive further advancements in our understanding of biological systems and improve clinical outcomes.

2.2 Training and Optimization Strategies ​

Deep learning has emerged as a transformative tool in various domains of biology, showcasing its versatility and efficacy in analyzing complex biological data. Its applications span across genomics, protein design, drug discovery, imaging, and more, leveraging vast datasets to uncover intricate patterns and relationships.

In genomics, deep learning has been instrumental in predicting the structure and function of genomic elements, such as promoters and enhancers, as well as gene expression levels. Liu et al. (2020) discuss the application of convolutional neural networks in genomics, highlighting how deep learning aids in understanding genomic data through enhanced predictive capabilities[2]. Moreover, deep learning techniques have been employed to analyze plant genome sequences, predict gene expression, and explore epigenetic features, showcasing its broad applicability in both animal and plant biology[4].

In the realm of protein design, deep learning has revolutionized the development of antibodies. Joubbi et al. (2024) illustrate how deep learning methods streamline the design and optimization processes, combining in vitro and in silico approaches to enhance antibody development against complex antigens[5]. This includes aspects such as design, folding, and affinity maturation, underscoring the significant advancements achieved through deep learning in the field of protein engineering.

Deep learning's impact extends to drug discovery, where it has shown superior performance compared to traditional machine learning algorithms. Techniques such as virtual screening have been enhanced by deep learning, enabling more efficient identification of potential drug candidates. Liu et al. (2019) developed a web server called DeepScreening that integrates state-of-the-art deep learning algorithms for virtual screening, allowing researchers to construct models and generate libraries for specific targets[13].

Imaging analysis is another critical area where deep learning has made significant contributions. Moen et al. (2019) highlight how deep learning algorithms have transformed cellular image analysis, enabling routine analyses that were previously challenging. Applications include image classification, segmentation, and object tracking, which are vital for interpreting complex biological images[14].

Furthermore, deep learning has been utilized in cancer research for diagnosis, prognosis, and treatment selection. Tran et al. (2021) provide an overview of emerging deep learning techniques applied to various omics data types, emphasizing the integration of genomic, methylation, and transcriptomic data to develop decision support tools in oncology[9]. This integration is crucial for understanding the complex biology of cancer and improving patient outcomes.

Despite these advancements, the application of deep learning in biology is not without challenges. Issues such as the need for large labeled datasets, interpretability of models, and the integration of multi-omics data present ongoing hurdles. As highlighted by Wainberg et al. (2018), while deep learning shows promise, ensuring the reliability and trustworthiness of these models remains a critical area for future research[15].

In summary, deep learning has established itself as a powerful methodology in biology, with applications spanning genomics, protein design, drug discovery, and imaging. Its ability to process and analyze vast amounts of complex data positions it as a cornerstone technology in advancing biological research and healthcare. The ongoing development of training and optimization strategies will be essential to address existing challenges and fully realize the potential of deep learning in this field.

3 Applications in Genomics ​

3.1 Genome Annotation and Variant Calling ​

Deep learning has emerged as a transformative tool in genomics, particularly in the areas of genome annotation and variant calling. The rapid advancements in genomic sequencing technologies have led to the generation of vast amounts of data, necessitating sophisticated analytical approaches. Deep learning techniques have been effectively utilized to enhance the accuracy and efficiency of these processes.

In genome annotation, deep learning models have been employed to predict various genomic features, such as promoters, enhancers, and gene expression levels. These models are capable of capturing complex patterns within high-dimensional genomic data, which traditional methods may struggle to identify. For instance, convolutional neural networks (CNNs) have been applied to analyze sequence data and predict regulatory elements, thus improving our understanding of gene regulation (Liu et al. 2020) [2].

Variant calling, the process of identifying genetic variants from sequencing data, has particularly benefited from deep learning methodologies. A notable example is the DeepVariant tool, which utilizes a deep convolutional neural network to analyze read pileups from next-generation sequencing data. This approach allows for the detection of single nucleotide polymorphisms (SNPs), insertions, deletions, and structural variants with higher accuracy than conventional methods. DeepVariant has demonstrated its ability to generalize across different genome builds and species, thereby enhancing variant calling across diverse datasets (Poplin et al. 2018) [16].

Moreover, deep learning models have been integrated into clinical genomics, where they assist in distinguishing true mutations from experimental errors, thus facilitating accurate diagnoses and prognostic evaluations. The ability to leverage biological, clinical, and laboratory variables in variant calling decisions has been shown to significantly improve the reliability of the results (Bohannan & Mitrofanova 2019) [17].

In summary, deep learning applications in genomics, specifically in genome annotation and variant calling, have led to significant improvements in the accuracy and efficiency of genomic analyses. These advancements not only enhance our understanding of genetic variations but also pave the way for precision medicine by enabling better-informed clinical decisions based on genomic data. The ongoing development of deep learning methodologies continues to promise further enhancements in genomic research and applications.

3.2 Gene Expression Analysis ​

Deep learning has emerged as a powerful tool in genomics, particularly in the analysis of gene expression. The advancements in genomic sequencing technologies have generated vast amounts of data, necessitating sophisticated analytical approaches to interpret complex biological information. Deep learning techniques have been successfully applied to various aspects of gene expression analysis, providing significant insights into the underlying biological processes.

One of the primary applications of deep learning in gene expression analysis is the prediction of gene expression levels based on genomic data. Liu et al. (2020) highlighted that deep learning has gained popularity in predicting the structure and function of genomic elements, including gene expression levels, through the use of convolutional neural networks (CNNs) and other architectures tailored for genomic data [2]. These models can capture intricate patterns within high-dimensional data that traditional machine learning methods may overlook.

Moreover, deep learning models have been utilized to integrate multi-omics data, which combines various layers of biological information, such as genetic, epigenetic, and transcriptomic data. For instance, Seal et al. (2020) developed a deep learning regression model that estimates gene expression from DNA methylation and copy number variation, showcasing the capability of deep learning to integrate diverse omics data and improve predictive accuracy [18]. Their model achieved an impressive accuracy of 95.1% in classifying disease states, illustrating the potential of deep learning in enhancing gene expression analysis.

Additionally, deep learning approaches have been employed to decipher the regulatory mechanisms governing gene expression. Sokolova et al. (2024) discussed the development of sequence-based deep learning models that link DNA patterns to biochemical properties, including modeling epigenetic marks and gene expression with tissue and cell-type specificity [19]. These models are instrumental in predicting the functional consequences of genetic variants, even those that are rare or previously unobserved, thus advancing our understanding of gene regulation.

Furthermore, deep learning's application extends to motif discovery and gene expression regulation. Jeong et al. (2025) reviewed various deep learning models utilized in genomics, emphasizing their role in motif finding and the regulation of gene expression [20]. The efficiency and reliability of these models have been demonstrated in real-world applications, providing a robust framework for genomic profile prediction.

In summary, deep learning has significantly advanced gene expression analysis in genomics through its ability to handle large datasets, integrate multi-omics information, and uncover complex regulatory patterns. The ongoing development and refinement of deep learning models promise to further enhance our understanding of gene expression and its implications in health and disease.

4 Applications in Proteomics ​

4.1 Protein Structure Prediction ​

Deep learning has emerged as a transformative force in the field of biology, particularly in proteomics and protein structure prediction. The application of deep learning methodologies has significantly enhanced our understanding of protein dynamics, interactions, and functions. Below are the key applications of deep learning in protein structure prediction, as detailed in recent literature.

  1. Protein Structure Prediction: Deep learning has revolutionized the prediction of protein structures from amino acid sequences. Techniques such as deep neural networks have demonstrated remarkable accuracy in predicting three-dimensional (3D) structural components of proteins. Notably, methods like AlphaFold2 have set new benchmarks by achieving near-experimental accuracy in protein structure prediction, effectively bridging the gap between sequence and structure [21][22].

  2. Contact Map Prediction: Deep learning models have been successfully employed to predict inter-residue contacts, which are crucial for understanding protein folding and stability. This has implications not only for predicting the structure of individual proteins but also for modeling protein-protein interactions and complexes [23][24].

  3. Modeling Protein-Protein Interactions: The ability to accurately predict how proteins interact with each other is vital for understanding cellular processes and disease mechanisms. Deep learning techniques have facilitated high-throughput predictions of protein-protein and protein-ligand interactions, thus advancing the field of drug discovery [25][26].

  4. Post-Translational Modifications (PTMs): Deep learning approaches are also being utilized to predict PTMs, which play significant roles in regulating protein function. These methods analyze vast datasets to identify patterns and predict modifications that may influence protein activity [27].

  5. De Novo Peptide Sequencing: In addition to predicting existing protein structures, deep learning has been applied to de novo peptide sequencing from mass spectrometry data. This allows for the identification of novel peptides and proteins, expanding our understanding of the proteome [27].

  6. Integration with Structural Information: Recent advances in deep learning have highlighted the importance of integrating structural data into predictive models. This integration enhances the accuracy of predictions and enables the modeling of complex protein assemblies and interactions [25][28].

  7. Evolutionary and Contact-Based Predictions: Deep learning has been instrumental in applying evolutionary data to enhance contact predictions, thus improving the accuracy of protein structure predictions. This approach leverages homology and co-evolutionary information to predict how protein sequences relate to their structures [29].

  8. Addressing Limitations and Future Directions: While deep learning has significantly advanced protein structure prediction, challenges remain, including data quality and the need for more robust models. Future directions include the development of more sophisticated architectures and methods that can handle the complexities of protein interactions and functions [30][31].

In summary, deep learning applications in protein structure prediction have fundamentally altered the landscape of proteomics, enabling unprecedented insights into protein dynamics and interactions. The integration of deep learning with traditional experimental methods promises to further enhance our understanding of biological systems and accelerate advancements in drug discovery and therapeutic design.

4.2 Protein-Protein Interaction Prediction ​

Deep learning has emerged as a transformative tool in the field of proteomics, particularly in the prediction of protein-protein interactions (PPIs), which are crucial for understanding various biological processes and disease mechanisms. The integration of deep learning techniques has significantly enhanced the accuracy and efficiency of PPI predictions compared to traditional methods.

One prominent application of deep learning in PPI prediction involves the modeling of protein complexes. Recent advances have shown that deep learning approaches can predict inter-protein contacts and model protein-protein complex structures effectively. For instance, methods such as those developed by Lin et al. (2024) utilize deep learning for end-to-end modeling of protein complexes, demonstrating improvements over conventional docking methods by directly predicting inter-protein contacts[23]. This capability is essential for elucidating the structural basis of protein interactions, which are often critical for cellular functions.

Additionally, the application of deep learning has enabled high-throughput screening of protein interactions on a proteome-wide scale. Zhang et al. (2025) reported a deep learning-based approach that successfully predicted 17,849 interactions in the human proteome, achieving a precision of 90% with a systematic screening of 200 million protein pairs. This study highlights the potential of deep learning to uncover previously unrecognized interactions, thereby expanding our understanding of protein functions and their implications in human diseases[32].

Deep learning models have also been developed to integrate various data sources for improved PPI predictions. For example, Tran et al. (2024) introduced a model that combines deep learning with feature fusion techniques, achieving accuracies of 96.34% and 99.30% on different datasets[33]. This integration of handcrafted features with protein sequence embeddings enhances the model's generalization capability across different biological contexts.

Moreover, the emergence of transformer-based architectures has further revolutionized PPI prediction. The DeepHomo2.0 model, developed by Lin et al. (2023), employs transformer-enhanced deep learning to predict contacts in homodimeric complexes, achieving significant precision improvements over existing methods[24]. Such advancements indicate the potential of deep learning to not only predict interactions but also to refine the understanding of the underlying mechanisms governing these interactions.

The challenges in PPI prediction, such as data quality and the need for accurate validation, are being addressed through the continuous evolution of deep learning methodologies. Recent reviews have summarized the state-of-the-art computational methods, highlighting the role of deep learning in overcoming limitations associated with traditional experimental approaches[34]. These developments underscore the growing ecosystem of deep learning methods aimed at modeling protein interactions, which is crucial for advancing drug discovery and therapeutic development.

In summary, deep learning has significantly enhanced the capabilities of PPI prediction through various innovative approaches, enabling high-accuracy predictions, the integration of diverse data sources, and the modeling of complex protein structures. These advancements hold promise for further elucidating the intricate networks of protein interactions that underlie biological functions and disease mechanisms.

5 Applications in Drug Discovery ​

5.1 Drug Target Identification ​

Deep learning has emerged as a transformative technology in the field of drug discovery, particularly in the area of drug target identification. This advancement is driven by the need for more efficient and accurate methods to discover novel drug targets, which are crucial for developing effective therapeutics. Traditional experimental methods for identifying druggable proteins are often expensive, time-consuming, and labor-intensive. In contrast, deep learning techniques provide computational approaches that can streamline and enhance this process.

One of the primary applications of deep learning in drug target identification is the development of predictive models that can accurately identify potential druggable proteins. For instance, Yu et al. (2022) developed a hybrid deep learning model that achieved an impressive accuracy of 90.0% on a benchmark dataset for predicting druggable proteins. This model utilized a combination of convolutional neural networks (CNNs) and recurrent neural networks (RNNs), demonstrating the effectiveness of integrating different architectures to improve prediction performance [35].

Moreover, Liu et al. (2019) introduced DeepScreening, a web server that employs state-of-the-art deep learning algorithms for virtual screening. This platform allows users to construct deep learning models that can utilize either public or user-provided datasets to perform virtual screenings for specific drug targets. The user-friendly interface facilitates the generation of target-focused de novo libraries, enabling researchers to identify potential chemical probes or drugs more efficiently [13].

The use of deep learning also extends to the prediction of drug-target interactions (DTIs), which is essential for understanding how drugs interact with their biological targets. The deep learning-based methods have shown superior performance in predicting these interactions compared to traditional machine learning approaches. For example, deep learning models can learn complex patterns from large datasets, thereby enhancing the accuracy of DTI predictions. This capability is critical for the identification of novel drug targets and understanding the dynamics of drug action [36].

Furthermore, the integration of deep learning with network-based approaches has been utilized to identify drug targets within heterogeneous biological networks. Zeng et al. (2020) developed deepDTnet, a deep learning methodology that leverages various chemical, genomic, and cellular network profiles to predict novel molecular targets for known drugs. This approach demonstrated high accuracy and has the potential to accelerate drug repurposing efforts [37].

In summary, deep learning has significantly advanced the field of drug target identification by providing robust predictive models that enhance the efficiency and accuracy of discovering druggable proteins and their interactions. These advancements are crucial for accelerating drug discovery processes and improving therapeutic outcomes.

5.2 Predictive Modeling of Drug Responses ​

Deep learning (DL) has emerged as a transformative tool in the field of drug discovery, particularly in the predictive modeling of drug responses. This approach leverages complex algorithms to analyze high-dimensional biological and chemical data, enabling the prediction of how cancer cell lines and patients respond to various anti-cancer treatments.

One of the primary applications of DL in drug response prediction is the development of models that can predict the sensitivity of tumors to specific anti-cancer treatments. Traditional methods often struggle with the intricacies of biological data, whereas DL algorithms, particularly those trained on high-throughput screening data, can effectively model these complexities. Recent studies have highlighted the potential of DL in this domain, showcasing its ability to predict responses based on both biological and chemical characteristics, which is critical for precision medicine (Baptista et al., 2021) [38].

The introduction of specialized DL libraries, such as DeepDR, has further facilitated drug response prediction by automating various processes involved in model construction, training, and inference. DeepDR incorporates multiple types of drug and cell features, allowing researchers to implement a wide array of DL models for predicting drug responses efficiently (Jiang & Li, 2024) [39].

Moreover, the integration of multi-omics data has been pivotal in enhancing the accuracy of drug response predictions. Deep learning models have been applied to analyze data from over 1000 cancer cell lines, leading to improved prediction capabilities. However, challenges remain, particularly in predicting responses to drugs not present in the training datasets, indicating the need for continuous refinement of these models (Chen & Zhang, 2022) [40].

In addition to predicting drug responses, DL has also been utilized in the identification of druggable proteins and drug-target interactions. The ability to rapidly and accurately identify potential drug targets is crucial for drug development, and DL algorithms have shown promise in this area by analyzing various protein features and employing different architectures to improve prediction accuracy (Yu et al., 2022) [35].

Overall, the applications of deep learning in predictive modeling of drug responses represent a significant advancement in drug discovery. By harnessing the power of DL, researchers can better understand and predict how different cancer therapies will perform, ultimately leading to more personalized and effective treatment strategies. As the field progresses, ongoing research will likely focus on addressing the limitations of current models and exploring new methodologies to enhance predictive accuracy and efficiency in drug response prediction.

6 Challenges and Limitations ​

6.1 Data Quality and Availability ​

Deep learning has emerged as a transformative technology in the field of biology, offering substantial applications across various domains, including genomics, protein structure prediction, and medical imaging. However, the successful implementation of deep learning in biological contexts is often impeded by challenges related to data quality and availability.

One significant challenge is the reliance on high-quality training datasets. In many biological applications, especially those involving deep learning, the performance of models is heavily contingent upon the quality of the training data. For instance, deep learning models typically require large-scale, well-annotated datasets to achieve satisfactory performance. However, obtaining such datasets can be both challenging and expensive, particularly in fields like plant disease recognition, where high-quality images and annotations are crucial for training effective models[41].

Additionally, the complexity and heterogeneity of biological data pose further obstacles. Biological datasets are often derived from various sources and may exhibit high dimensionality and variability, which complicates their analysis. For example, deep learning techniques have shown promise in extracting features from large and complex datasets, yet they also require robust data preprocessing to ensure that the models can learn effectively from the underlying patterns[9][42].

Moreover, there are legal and privacy constraints associated with handling sensitive health records, which limit the availability of high-quality data necessary for training deep learning models in medical applications. These constraints can hinder the development of predictive models that could otherwise provide valuable insights into patient care and treatment[1].

Another notable limitation is the interpretability of deep learning models. While these models can achieve high accuracy, understanding how they arrive at specific predictions remains a significant challenge. This lack of interpretability can impede their acceptance in clinical settings, where practitioners require explanations for the decisions made by AI systems[8][43].

In summary, while deep learning holds considerable promise for advancing biological research and applications, the challenges related to data quality and availability must be addressed. Ensuring access to high-quality, well-annotated datasets, improving the interpretability of models, and navigating legal constraints are essential steps for the successful integration of deep learning into biological and medical fields.

6.2 Interpretability of Deep Learning Models ​

Deep learning has emerged as a transformative approach in the field of biology, with applications spanning various domains, including genomics, cancer diagnosis, and vaccine design. However, the adoption of deep learning in biological contexts is accompanied by significant challenges, particularly concerning the interpretability of models.

In the realm of genomics and epigenomics, deep learning models have been employed to analyze complex biological data, enabling researchers to identify patterns and make predictions regarding genetic variations and their effects on cellular processes. These models excel at processing high-dimensional datasets and have shown promise in tasks such as sequence motif identification and gene expression analysis. Despite these advancements, the black-box nature of deep neural networks (DNNs) poses a challenge, as their predictions often lack transparency, making it difficult for researchers to understand the underlying mechanisms driving these predictions[44].

In oncology, deep learning has been utilized for cancer diagnosis, prognosis, and treatment selection. Researchers have developed models that integrate various omics data types, such as genomic and transcriptomic data, to improve decision-making in cancer care. However, the need for more explainable models remains a significant limitation. Many current models provide limited insight into how predictions are made, hindering their clinical application and acceptance by healthcare professionals[9].

The interpretability of deep learning models is critical for establishing trust and ensuring their efficacy in healthcare applications. Several strategies have been developed to enhance interpretability, such as designing models that can output rationales for their predictions. This is particularly important in biomedicine, where stakeholders require a clear understanding of how models arrive at their conclusions[15].

Recent research has highlighted the necessity of integrating biological knowledge into deep learning frameworks to enhance both performance and interpretability. For instance, incorporating prior biological relational and network knowledge can improve model generalization and facilitate a better understanding of the biological mechanisms involved[45]. This concept of bio-centric interpretability aims to create models that not only perform well but also provide insights that are meaningful within the biological context[46].

Despite the promising advancements in deep learning applications across various biological fields, the ongoing challenges related to interpretability remain a barrier to their widespread clinical adoption. Future research is needed to develop more transparent models that can effectively communicate their decision-making processes, thereby bridging the gap between computational power and biological understanding[47].

7 Future Directions ​

7.1 Integration with Other Computational Methods ​

Deep learning has emerged as a transformative tool in biology, particularly in its ability to handle large and complex datasets, thereby enhancing various research and clinical applications. The applications of deep learning in biology are extensive, encompassing several key areas such as genomics, drug discovery, patient classification, and cellular image analysis.

In genomics, deep learning has been instrumental in predicting the structure and function of genomic elements, such as promoters and enhancers, as well as gene expression levels. For instance, Liu et al. (2020) discussed the use of convolutional neural networks to derive biological insights from genomic data, emphasizing the dramatic improvements in performance over conventional machine learning techniques[2]. Additionally, deep learning has shown promise in the analysis of omics data types, including genomic, methylation, and transcriptomic data, facilitating cancer diagnosis and treatment selection[9].

In the realm of drug discovery, deep learning has been applied to bioactivity predictions, molecular design, and synthesis predictions, significantly streamlining the drug development process[6]. The technology has also been leveraged for the analysis of complex biological images, transforming cellular image analysis through improved classification, segmentation, and object tracking capabilities[14].

Furthermore, deep learning is increasingly being integrated with other computational methods, enhancing its efficacy and application scope. For example, machine learning techniques are being combined with bioinformatics frameworks to improve feature extraction and model validation, thus addressing challenges in data interpretation and analysis[48]. The integration of deep learning with traditional computational methods allows for more comprehensive analyses of biological systems, enabling researchers to uncover hidden patterns and make more accurate predictions.

Future directions for deep learning in biology include further integration with wet-bench experimentation, which may help elucidate the underlying principles of cellular functionalities and enhance the interpretability of models. This approach aims to create an integrated framework where computational predictions can inform experimental designs and vice versa[49]. Additionally, the exploration of multi-modal data fusion, where different types of biological data are analyzed together, presents a promising avenue for improving diagnostic precision and treatment personalization in clinical settings[50].

Overall, the applications of deep learning in biology are rapidly expanding, and its integration with other computational methods is likely to drive significant advancements in understanding complex biological processes and improving clinical outcomes.

7.2 Ethical Considerations in Biological Research ​

Deep learning has emerged as a transformative tool in the field of biology, offering innovative applications across various domains. Its ability to analyze complex, high-dimensional data makes it particularly suited for biological research, which often involves large datasets generated from genomics, proteomics, and imaging studies. The applications of deep learning in biology are extensive and include the following key areas:

  1. Genomics: Deep learning techniques have been widely utilized to predict gene expression levels, identify regulatory elements such as promoters and enhancers, and analyze genomic sequences for functional components. The capacity of deep learning to uncover intricate patterns in genomic data holds significant promise for advancing our understanding of genetic regulation and disease mechanisms (Liu et al., 2020; Wang et al., 2025).

  2. Cancer Diagnosis and Treatment: In oncology, deep learning has shown potential in analyzing omics data, including genomic, transcriptomic, and methylation data, to enhance cancer diagnosis, prognosis, and treatment selection. By integrating diverse data types, deep learning models can support decision-making in precision oncology, although challenges such as the need for more explainable models and phenotypically rich datasets remain (Tran et al., 2021).

  3. Protein Design and Antibody Development: The application of deep learning in protein structure prediction and antibody design is gaining traction. Novel techniques have been developed to streamline the design and optimization of antibodies, facilitating the generation of lead candidates against complex antigens (Joubbi et al., 2024). This area showcases how deep learning can expedite the development of biologics, a crucial aspect of therapeutic interventions.

  4. Cellular Image Analysis: Deep learning algorithms are transforming the analysis of biological images, enabling tasks such as image classification, segmentation, and object tracking. These capabilities are crucial for understanding cellular behaviors and interactions, which are often difficult to quantify using traditional methods (Moen et al., 2019).

  5. Biological Networks: The integration of deep learning with biological network analysis allows for the extraction of meaningful insights from complex biomolecular interaction data. By leveraging deep learning, researchers can better understand the underlying biological systems and discover therapeutic targets for complex diseases (Jin et al., 2021).

  6. Synthetic Biology and Drug Discovery: In synthetic biology, deep learning is being employed to design and optimize genetic circuits and pathways. Additionally, its application in drug discovery has expanded beyond bioactivity predictions to include de novo molecular design and synthesis prediction, demonstrating the versatility of deep learning in pharmaceutical research (Chen et al., 2018).

As the field progresses, several future directions for deep learning in biology can be anticipated. These include the development of more sophisticated algorithms capable of integrating multi-omics data, enhancing model interpretability, and addressing the ethical considerations associated with data privacy and the use of sensitive health information. The ability to effectively communicate the results of deep learning analyses to a broader audience, including non-experts, will also be crucial for fostering collaboration between computational and experimental biologists.

Ethical considerations in biological research utilizing deep learning are paramount, particularly concerning data privacy, informed consent, and the potential for bias in model training. As deep learning continues to influence biological research, it is essential to establish guidelines that ensure ethical practices while maximizing the benefits of this powerful technology (Ching et al., 2018).

In summary, deep learning is revolutionizing various aspects of biology, from genomics and cancer research to protein design and imaging analysis. As researchers navigate the challenges and ethical implications of these advancements, the potential for deep learning to significantly impact biological research and healthcare remains substantial.

8 Conclusion ​

Deep learning has significantly transformed the landscape of biological research, offering innovative solutions across various domains such as genomics, proteomics, drug discovery, and medical imaging. Key findings indicate that deep learning methodologies have greatly enhanced the accuracy and efficiency of genomic analyses, protein structure predictions, and drug response modeling. However, challenges such as data quality, model interpretability, and ethical considerations remain critical hurdles that must be addressed. Future research directions should focus on integrating deep learning with traditional computational methods, improving model transparency, and navigating ethical implications in biological research. As the field evolves, the potential for deep learning to further revolutionize our understanding of complex biological systems and improve clinical outcomes continues to expand.

References ​

  • [1] Travers Ching;Daniel S Himmelstein;Brett K Beaulieu-Jones;Alexandr A Kalinin;Brian T Do;Gregory P Way;Enrico Ferrero;Paul-Michael Agapow;Michael Zietz;Michael M Hoffman;Wei Xie;Gail L Rosen;Benjamin J Lengerich;Johnny Israeli;Jack Lanchantin;Stephen Woloszynek;Anne E Carpenter;Avanti Shrikumar;Jinbo Xu;Evan M Cofer;Christopher A Lavender;Srinivas C Turaga;Amr M Alexandari;Zhiyong Lu;David J Harris;Dave DeCaprio;Yanjun Qi;Anshul Kundaje;Yifan Peng;Laura K Wiley;Marwin H S Segler;Simina M Boca;S Joshua Swamidass;Austin Huang;Anthony Gitter;Casey S Greene. Opportunities and obstacles for deep learning in biology and medicine.. Journal of the Royal Society, Interface(IF=3.5). 2018. PMID:29618526. DOI: 10.1098/rsif.2017.0387.
  • [2] Jianxiao Liu;Jiying Li;Hai Wang;Jianbing Yan. Application of deep learning in genomics.. Science China. Life sciences(IF=9.5). 2020. PMID:33051704. DOI: 10.1007/s11427-020-1804-5.
  • [3] . Deep learning for genomics.. Nature genetics(IF=29.0). 2019. PMID:30578416. DOI: 10.1038/s41588-018-0328-0.
  • [4] Zhenye Wang;Hao Yuan;Jianbing Yan;Jianxiao Liu. Identification, characterization, and design of plant genome sequences using deep learning.. The Plant journal : for cell and molecular biology(IF=5.7). 2025. PMID:39666835. DOI: 10.1111/tpj.17190.
  • [5] Sara Joubbi;Alessio Micheli;Paolo Milazzo;Giuseppe Maccari;Giorgio Ciano;Dario Cardamone;Duccio Medini. Antibody design using deep learning: from sequence and structure design to affinity maturation.. Briefings in bioinformatics(IF=7.7). 2024. PMID:38960409. DOI: 10.1093/bib/bbae307.
  • [6] Hongming Chen;Ola Engkvist;Yinhai Wang;Marcus Olivecrona;Thomas Blaschke. The rise of deep learning in drug discovery.. Drug discovery today(IF=7.5). 2018. PMID:29366762. DOI: 10.1016/j.drudis.2018.01.039.
  • [7] Carolin A Rickert;Oliver Lieleg. Machine learning approaches for biomolecular, biophysical, and biomaterials research.. Biophysics reviews(IF=3.4). 2022. PMID:38505413. DOI: 10.1063/5.0082179.
  • [8] Chensi Cao;Feng Liu;Hai Tan;Deshou Song;Wenjie Shu;Weizhong Li;Yiming Zhou;Xiaochen Bo;Zhi Xie. Deep Learning and Its Applications in Biomedicine.. Genomics, proteomics & bioinformatics(IF=7.9). 2018. PMID:29522900. DOI: 10.1016/j.gpb.2017.07.003.
  • [9] Khoa A Tran;Olga Kondrashova;Andrew Bradley;Elizabeth D Williams;John V Pearson;Nicola Waddell. Deep learning in cancer diagnosis, prognosis and treatment selection.. Genome medicine(IF=11.2). 2021. PMID:34579788. DOI: 10.1186/s13073-021-00968-x.
  • [10] Erik Meijering. A bird's-eye view of deep learning in bioimage analysis.. Computational and structural biotechnology journal(IF=4.1). 2020. PMID:32994890. DOI: 10.1016/j.csbj.2020.08.003.
  • [11] Christof Angermueller;Tanel Pärnamaa;Leopold Parts;Oliver Stegle. Deep learning for computational biology.. Molecular systems biology(IF=7.7). 2016. PMID:27474269. DOI: 10.15252/msb.20156651.
  • [12] Mufti Mahmud;Mohammed Shamim Kaiser;Amir Hussain;Stefano Vassanelli. Applications of Deep Learning and Reinforcement Learning to Biological Data.. IEEE transactions on neural networks and learning systems(IF=8.9). 2018. PMID:29771663. DOI: 10.1109/TNNLS.2018.2790388.
  • [13] Zhihong Liu;Jiewen Du;Jiansong Fang;Yulong Yin;Guohuan Xu;Liwei Xie. DeepScreening: a deep learning-based screening web server for accelerating drug discovery.. Database : the journal of biological databases and curation(IF=3.6). 2019. PMID:31608949. DOI: 10.1093/database/baz104.
  • [14] Erick Moen;Dylan Bannon;Takamasa Kudo;William Graf;Markus Covert;David Van Valen. Deep learning for cellular image analysis.. Nature methods(IF=32.1). 2019. PMID:31133758. DOI: 10.1038/s41592-019-0403-1.
  • [15] Michael Wainberg;Daniele Merico;Andrew Delong;Brendan J Frey. Deep learning in biomedicine.. Nature biotechnology(IF=41.7). 2018. PMID:30188539. DOI: 10.1038/nbt.4233.
  • [16] Ryan Poplin;Pi-Chuan Chang;David Alexander;Scott Schwartz;Thomas Colthurst;Alexander Ku;Dan Newburger;Jojo Dijamco;Nam Nguyen;Pegah T Afshar;Sam S Gross;Lizzie Dorfman;Cory Y McLean;Mark A DePristo. A universal SNP and small-indel variant caller using deep neural networks.. Nature biotechnology(IF=41.7). 2018. PMID:30247488. DOI: 10.1038/nbt.4235.
  • [17] Zachary S Bohannan;Antonina Mitrofanova. Calling Variants in the Clinic: Informed Variant Calling Decisions Based on Biological, Clinical, and Laboratory Variables.. Computational and structural biotechnology journal(IF=4.1). 2019. PMID:31049166. DOI: 10.1016/j.csbj.2019.04.002.
  • [18] Dibyendu Bikash Seal;Vivek Das;Saptarsi Goswami;Rajat K De. Estimating gene expression from DNA methylation and copy number variation: A deep learning regression model for multi-omics integration.. Genomics(IF=3.0). 2020. PMID:32234433. DOI: 10.1016/j.ygeno.2020.03.021.
  • [19] Ksenia Sokolova;Kathleen M Chen;Yun Hao;Jian Zhou;Olga G Troyanskaya. Deep Learning Sequence Models for Transcriptional Regulation.. Annual review of genomics and human genetics(IF=7.9). 2024. PMID:38594933. DOI: 10.1146/annurev-genom-021623-024727.
  • [20] Chang Beom Jeong;Hyein Cho;Daechan Park. Deep learning application for genomic data analysis.. BMB reports(IF=3.3). 2025. PMID:40962325. DOI: .
  • [21] Dongjin Lee;Dapeng Xiong;Shayne Wierbowski;Le Li;Siqi Liang;Haiyuan Yu. Deep learning methods for 3D structural proteome and interactome modeling.. Current opinion in structural biology(IF=7.0). 2022. PMID:35139457. DOI: 10.1016/j.sbi.2022.102329.
  • [22] Jue Wang;Joseph L Watson;Sidney L Lisanza. Protein Design Using Structure-Prediction Networks: AlphaFold and RoseTTAFold as Protein Structure Foundation Models.. Cold Spring Harbor perspectives in biology(IF=8.4). 2024. PMID:38438190. DOI: 10.1101/cshperspect.a041472.
  • [23] Peicong Lin;Hao Li;Sheng-You Huang. Deep learning in modeling protein complex structures: From contact prediction to end-to-end approaches.. Current opinion in structural biology(IF=7.0). 2024. PMID:38402744. DOI: 10.1016/j.sbi.2024.102789.
  • [24] Peicong Lin;Yumeng Yan;Sheng-You Huang. DeepHomo2.0: improved protein-protein contact prediction of homodimers by transformer-enhanced deep learning.. Briefings in bioinformatics(IF=7.7). 2023. PMID:36440949. DOI: 10.1093/bib/bbac499.
  • [25] Min Su Yoon;Byunghyun Bae;Kunhee Kim;Hahnbeom Park;Minkyung Baek. Deep learning methods for proteome-scale interaction prediction.. Current opinion in structural biology(IF=7.0). 2025. PMID:39848140. DOI: 10.1016/j.sbi.2024.102981.
  • [26] Yajie Meng;Zhuang Zhang;Chang Zhou;Xianfang Tang;Xinrong Hu;Geng Tian;Jialiang Yang;Yuhua Yao. Protein structure prediction via deep learning: an in-depth review.. Frontiers in pharmacology(IF=4.8). 2025. PMID:40248099. DOI: 10.3389/fphar.2025.1498662.
  • [27] Bo Wen;Wen-Feng Zeng;Yuxing Liao;Zhiao Shi;Sara R Savage;Wen Jiang;Bing Zhang. Deep Learning in Proteomics.. Proteomics(IF=3.9). 2020. PMID:32939979. DOI: 10.1002/pmic.201900335.
  • [28] Ben Shor;Dina Schneidman-Duhovny. Integrative modeling meets deep learning: Recent advances in modeling protein assemblies.. Current opinion in structural biology(IF=7.0). 2024. PMID:38795564. DOI: 10.1016/j.sbi.2024.102841.
  • [29] Donghyuk Suh;Jai Woo Lee;Sun Choi;Yoonji Lee. Recent Applications of Deep Learning Methods on Evolution- and Contact-Based Protein Structure Prediction.. International journal of molecular sciences(IF=4.9). 2021. PMID:34199677. DOI: 10.3390/ijms22116032.
  • [30] Nicolae Sapoval;Amirali Aghazadeh;Michael G Nute;Dinler A Antunes;Advait Balaji;Richard Baraniuk;C J Barberan;Ruth Dannenfelser;Chen Dun;Mohammadamin Edrisi;R A Leo Elworth;Bryce Kille;Anastasios Kyrillidis;Luay Nakhleh;Cameron R Wolfe;Zhi Yan;Vicky Yao;Todd J Treangen. Current progress and open challenges for applying deep learning across the biosciences.. Nature communications(IF=15.7). 2022. PMID:35365602. DOI: 10.1038/s41467-022-29268-7.
  • [31] Saber Saharkhiz;Mehrnaz Mostafavi;Amin Birashk;Shiva Karimian;Shayan Khalilollah;Sohrab Jaferian;Yalda Yazdani;Iraj Alipourfard;Yun Suk Huh;Marzieh Ramezani Farani;Reza Akhavan-Sigari. The State-of-the-Art Overview to Application of Deep Learning in Accurate Protein Design and Structure Prediction.. Topics in current chemistry (Cham)(IF=8.8). 2024. PMID:38965117. DOI: 10.1007/s41061-024-00469-6.
  • [32] Jing Zhang;Ian R Humphreys;Jimin Pei;Jinuk Kim;Chulwon Choi;Rongqing Yuan;Jesse Durham;Siqi Liu;Hee-Jung Choi;Minkyung Baek;David Baker;Qian Cong. Predicting protein-protein interactions in the human proteome.. Science (New York, N.Y.)(IF=45.8). 2025. PMID:40997207. DOI: 10.1126/science.adt1630.
  • [33] Hoai-Nhan Tran;Phuc-Xuan-Quynh Nguyen;Fei Guo;Jianxin Wang. Prediction of Protein-Protein Interactions Based on Integrating Deep Learning and Feature Fusion.. International journal of molecular sciences(IF=4.9). 2024. PMID:38892007. DOI: 10.3390/ijms25115820.
  • [34] Neal Kewalramani;Andrew Emili;Mark Crovella. State-of-the-art computational methods to predict protein-protein interactions with high accuracy and coverage.. Proteomics(IF=3.9). 2023. PMID:37401192. DOI: 10.1002/pmic.202200292.
  • [35] Lezheng Yu;Li Xue;Fengjuan Liu;Yizhou Li;Runyu Jing;Jiesi Luo. The applications of deep learning algorithms on in silico druggable proteins identification.. Journal of advanced research(IF=13.0). 2022. PMID:36328750. DOI: 10.1016/j.jare.2022.01.009.
  • [36] Xin Zeng;Shu-Juan Li;Shuang-Qing Lv;Meng-Liang Wen;Yi Li. A comprehensive review of the recent advances on predicting drug-target affinity based on deep learning.. Frontiers in pharmacology(IF=4.8). 2024. PMID:38628639. DOI: 10.3389/fphar.2024.1375522.
  • [37] Xiangxiang Zeng;Siyi Zhu;Weiqiang Lu;Zehui Liu;Jin Huang;Yadi Zhou;Jiansong Fang;Yin Huang;Huimin Guo;Lang Li;Bruce D Trapp;Ruth Nussinov;Charis Eng;Joseph Loscalzo;Feixiong Cheng. Target identification among known drugs by deep learning from heterogeneous networks.. Chemical science(IF=7.4). 2020. PMID:34123272. DOI: 10.1039/c9sc04336e.
  • [38] Delora Baptista;Pedro G Ferreira;Miguel Rocha. Deep learning for drug response prediction in cancer.. Briefings in bioinformatics(IF=7.7). 2021. PMID:31950132. DOI: 10.1093/bib/bbz171.
  • [39] Zhengxiang Jiang;Pengyong Li. DeepDR: a deep learning library for drug response prediction.. Bioinformatics (Oxford, England)(IF=5.4). 2024. PMID:39558584. DOI: 10.1093/bioinformatics/btae688.
  • [40] Yurui Chen;Louxin Zhang. How much can deep learning improve prediction of the responses to drugs in cancer cell lines?. Briefings in bioinformatics(IF=7.7). 2022. PMID:34529029. DOI: 10.1093/bib/bbab378.
  • [41] Mingle Xu;Hyongsuk Kim;Jucheng Yang;Alvaro Fuentes;Yao Meng;Sook Yoon;Taehyun Kim;Dong Sun Park. Embracing limited and imperfect training datasets: opportunities and challenges in plant disease recognition using deep learning.. Frontiers in plant science(IF=4.8). 2023. PMID:37810377. DOI: 10.3389/fpls.2023.1225409.
  • [42] Yu Li;Chao Huang;Lizhong Ding;Zhongxiao Li;Yijie Pan;Xin Gao. Deep learning in bioinformatics: Introduction, application, and perspective in the big data era.. Methods (San Diego, Calif.)(IF=4.3). 2019. PMID:31022451. DOI: 10.1016/j.ymeth.2019.04.008.
  • [43] Hao He;Sen Yan;Danya Lyu;Mengxi Xu;Ruiqian Ye;Peng Zheng;Xinyu Lu;Lei Wang;Bin Ren. Deep Learning for Biospectroscopy and Biospectral Imaging: State-of-the-Art and Perspectives.. Analytical chemistry(IF=6.7). 2021. PMID:33599125. DOI: 10.1021/acs.analchem.0c04671.
  • [44] Amlan Talukder;Clayton Barham;Xiaoman Li;Haiyan Hu. Interpretation of deep learning in genomics and epigenomics.. Briefings in bioinformatics(IF=7.7). 2021. PMID:34020542. DOI: 10.1093/bib/bbaa177.
  • [45] Magdalena Wysocka;Oskar Wysocki;Marie Zufferey;Dónal Landers;André Freitas. A systematic review of biologically-informed deep learning models for cancer: fundamental trends for encoding and interpreting oncology data.. BMC bioinformatics(IF=3.3). 2023. PMID:37189058. DOI: 10.1186/s12859-023-05262-8.
  • [46] Wolfgang Esser-Skala;Nikolaus Fortelny. Reliable interpretability of biology-inspired deep neural networks.. NPJ systems biology and applications(IF=3.5). 2023. PMID:37816807. DOI: 10.1038/s41540-023-00310-8.
  • [47] Qiaoying Teng;Zhe Liu;Yuqing Song;Kai Han;Yang Lu. A survey on the interpretability of deep learning in medical diagnosis.. Multimedia systems(IF=3.1). 2022. PMID:35789785. DOI: 10.1007/s00530-022-00960-4.
  • [48] Noam Auslander;Ayal B Gussow;Eugene V Koonin. Incorporating Machine Learning into Established Bioinformatics Frameworks.. International journal of molecular sciences(IF=4.9). 2021. PMID:33809353. DOI: 10.3390/ijms22062903.
  • [49] Xiao Fu;Paul A Bates. Application of deep learning methods: From molecular modelling to patient classification.. Experimental cell research(IF=3.5). 2022. PMID:35810775. DOI: 10.1016/j.yexcr.2022.113278.
  • [50] Junwei Liu 刘俊伟;Xiaoping Cen 岑萧萍;Chenxin Yi 伊晨昕;Feng-Ao Wang 王烽傲;Junxiang Ding 丁俊翔;Jinyu Cheng 程瑾瑜;Qinhua Wu 吴沁桦;Baowen Gai 盖宝文;Yiwen Zhou 周奕雯;Ruikun He 贺瑞坤;Feng Gao 高峰;Yixue Li 李亦学. Challenges in AI-driven Biomedical Multimodal Data Fusion and Analysis.. Genomics, proteomics & bioinformatics(IF=7.9). 2025. PMID:40036568. DOI: 10.1093/gpbjnl/qzaf011.

MaltSci Intelligent Research Services ​

Search for more papers on MaltSci.com ​

Deep Learning · Biology · Genomics · Proteomics · Drug Discovery

© 2025 MaltSci