scRNA-seq offers a deeper understanding of cellular heterogeneity compared with traditional bulk cell analysis (50). It provides insights into gene expression at the single-cell level. The process involves isolating and lysing individual cells, followed by reverse transcribing mRNA and then amplifying it (50). However, previous single-cell isolation methods, such as manual picking (31,51–53), FACS-sorting (54–56) and integrated microfluidic circuits (57–59), have limitations in scalability due to cost, time and labor constraints. To address these limitations, advancements have been made to enhance efficiency and reduce costs. For example, Seq-Well (60), developed in 2017, is a portable and straightforward platform for massively-parallel scRNA-seq. Its working principle involves confining mRNA capture beads with unique barcodes and cells within small pores (pinholes), followed by sealing with a semipermeable membrane. This setup is conducive to efficient cell lysis and mRNA capture. After lysis, the beads are removed for parallel sequencing. The unique barcode on each bead allows for the identification of the originating cell of each transcript. Due to its simplicity and portability, Seq-Well can be implemented in a variety of settings, making it a versatile tool for single-cell genomics and transcriptomics research. It uses selective chemical functionalization (mRNA capture beads) to enable rapid cell lysis and efficient transcript capture while minimizing cross-contamination. Another scalable method, split-pool ligation-based transcriptome sequencing (SPLiT-seq), was proposed in 2018. SPLiT-seq enables efficient sample multiplexing without the need for specialized equipment, and is compatible with fixed cells or nuclei (61). In 2019, massively parallel RNA single cell sequencing (version 2.0) was introduced, combining sub-microliter reaction volumes, optimized enzymatic mixtures and an enhanced analytical pipeline (62). These methods substantially reduced costs, improved reproducibility and decreased well-to-well contamination (62). Unlike most single-cell transcriptomic profiling methods that focus on the 3′-end of polyadenylated transcripts, C1 Cap Analysis of Gene Expression, developed in 2019, detects transcript 5′-ends using an original sample multiplexing strategy in the C1TM microfluidic system. Analyzing transcript 5′-ends enhances the understanding of gene expression by allowing precise mapping of transcription start sites, which sheds light on the complexity of promoter usage and regulatory mechanisms. This approach not only reveals the diversity of transcript isoforms, contributing to protein variability, but also improves the accuracy of gene expression profiling across different conditions and cell types. Consequently, focusing on 5′-ends is crucial for unraveling the intricacies of gene regulation and the functional diversity within cellular processes (63).

Single-cell genome sequencing has allowed greater examination of genetic diversity, making it easier to analyze both de novo germline and somatic mutations in both normal and cancerous cells (23,64). In 2001, a simple method using rolling circle amplification was introduced for the amplification of vector DNA from single colonies or plaques, removing the requirement for lengthy growth periods and conventional DNA isolation techniques (65). In 2011, Navin et al (46) used flow-sorted nuclei, whole genome amplification and next-generation sequencing to precisely measure genomic copy numbers within individual nuclei. This method was used to explore the population structure and evolutionary dynamics of tumors in human breast cancer cases. Another method, multiple annealing and looping based amplification cycles, introduced quasi-linear preamplification in 2012 (66), reducing biases associated with nonlinear amplification (whole genome amplification). Furthermore, linear amplification via transposon insertion, proposed in 2017 (67), overcame limitations of current whole genome amplification methods, including amplification bias, amplification errors and limited resolution for detecting variations, enabling micro-copy number variation detection with kilobase resolution, while minimizing amplification biases and errors. Multiplexed end-tagging amplification of complementary strands (METACs) (68), developed in 2021, improved single-cell whole-genome amplification by leveraging the complementary strands of double-stranded DNA to filter out false positives and reduce sequencing costs, achieving high accuracy in detecting single-nucleotide variations and other genomic variants, which improves single-cell whole-genome amplification by leveraging both strands of the DNA. Unique tags are added to DNA ends before amplification, allowing for the pairing of complementary strands during sequencing. This method helps filter out false positives and reduces sequencing costs by requiring less sequencing depth for high accuracy. METACs is particularly effective in detecting single-nucleotide variations and other genomic variants, differing from traditional methods that often amplify only one DNA strand and may have higher error rates. It is applicable in fields such as single-cell genomics, clinical diagnostics and population genetics (68).

Single-cell epigenome analysis provides valuable insights into DNA methylation, histone modification and chromatin states, which influence cellular activity (69). In 2013, the first single-cell method for methylome analysis, single-cell reduced representation bisulfite sequencing, was introduced (70,71). This technique enables the measurement of the methylation state in ~10% of CpG sites through enrichment of CpG dense regions (70,71). These sites predominantly cover most promoters, yet a limitation is their relatively poor coverage of a number of crucial regulatory regions, such as enhancers (70). The post-bisulfite adapter tagging method involves bisulfite conversion prior to library preparation, ensuring that DNA degradation does not compromise the adapter-tagged fragments. This allows for the measurement of methylation at up to 50% of CpG sites in individual cells (53). Chromatin immunoprecipitation (ChIP) followed by sequencing demonstrates improved data compared with chromatin immunoprecipitation combined with DNA microarray (ChIP combined with DNA microarray technology is a method used to identify DNA regions that interact with specific proteins, by precipitating protein-DNA complexes and hybridizing the extracted DNA onto microarrays) by providing higher resolution, greater genomic coverage, increased sensitivity and cost-effectiveness, leading to more precise and comprehensive analysis of DNA-protein interactions, allowing genome-wide profiling of DNA-binding proteins, histone modifications and nucleosomes (72,73). Another method, single cell assay for transposase-accessible chromatin with sequencing, utilizes a transposase enzyme to insert sequencing adapters into open chromatin regions, revealing which parts of the genome are active or accessible in each cell. This technology is widely applied in epigenetics to understand cell-to-cell variability, identify regulatory elements such as enhancers and promoters, and explore the mechanisms of gene regulation in diverse cell types (57).

Single-cell protein mass spectrometry allows for comprehensive measurement of protein expression patterns in a cell (74). Cytometry by time of flight (75), a mass cytometry-based method, has been used to analyze surface and intracellular proteins using metal-labeled antibodies labeled with heavy metal isotopes, allowing simultaneous detection of multiple proteins in cells with minimal overlap and higher precision compared with fluorescent labels used in traditional flow cytometry. SCoPE by mass spectrometry (76), a high-throughput method, was developed based on liquid chromatography-tandem mass spectrometry techniques, and is a high-throughput method for single-cell proteomic analysis that allows for the isolation, enzymatic digestion, and subsequent identification and quantification of proteins from individual cells. This technique provides detailed insights into the protein composition of single cells, revealing cellular functions and heterogeneity at the proteomic level. Subsequent improvements led to the development of SCoPE2 (77), which offers enhanced quantitative accuracy, proteome coverage, sample preparation ease and cost-effectiveness.

Single-cell CRISPR sequencing is a cutting-edge technique that integrates CRISPR-Cas9, a powerful gene-editing tool, with SCS methods. This innovative approach primarily focuses on executing targeted gene edits at the single-cell level and then analyzing the consequent changes in the cell transcriptome using SCS, thereby allowing researchers to directly observe the effects of specific genetic alterations on gene expression in individual cells (78). Through this approach, a deeper understanding of gene functions, cellular networks and disease mechanisms is attainable. These techniques allow the use of compiled CRISPR libraries for collective cellular interventions, followed by high-throughput phenotypic analysis by using collective cellular interventions via CRISPR libraries to simultaneously edit multiple genes, followed by high-throughput phenotypic analysis, allowing for a comprehensive study of the resulting changes in cellular behavior and characteristics, revealing complex gene functions and interactions within cellular networks (78). To date, ≥30 different single-cell CRISPR techniques have been developed; the present review focuses on introducing a few representative technologies. Perturb-seq (79) and CRISP-seq (80) were among the first single-cell CRISPR techniques to be developed. Perturb-seq combines CRISPR-mediated gene perturbation with scRNA-seq for large-scale gene function screening and studying the impact of gene expression changes on cellular states (79). CRISP-seq, similar to Perturb-seq, focuses on studying individual genes or a small numbers of genes, assessing how specific gene-editing events affect cell function (80). By introducing specific gene alterations via CRISPR, CRISPR droplet sequencing (81), followed by scRNA-seq evaluates the influence of gene perturbations on cellular states and behaviors, which combines CRISPR technology with droplet-based single-cell sequencing, allowing simultaneous editing and gene expression profiling in individual cells. Unlike traditional CRISPR techniques that focus primarily on gene editing, CRISPR droplet sequencing integrates gene editing with detailed, single-cell level transcriptomic analysis, revealing how edits affect cellular functions (78). Mosaic-seq (82) generates cellular mosaics (a collection of cells in which each cell has a distinct genetic alteration) with various genetic perturbations, and analyzes the combined effects of these disruptions using scRNA-seq. This method is used to study the interactions between different genes and their impact on cellular functions. Direct-capture Perturb-seq (83), a variant of Perturb-seq, improves data quality and analysis efficiency by directly capturing and sequencing CRISPR-guided RNA, facilitating a more precise association between gene-editing events (deliberate alterations made to the genome using CRISPR-Cas9 technology, such as knocking out, knocking in or modifying specific gene sequences) and transcriptomic changes.

Single-cell CRISPR sequencing and its derivative technologies hold potential in cancer research. These techniques assess the complex molecular networks within tumor cells and aid the understanding of the TME, drug responses and mechanisms of treatment resistance (79). For instance, Jun et al (84), using in vitro experiments, explored all cytosine-to-thymine mutations in the exon regions of three genes (MAP2K1, KRAS and NRAS), revealing the insertions and deletions and transcriptomic markers contributing to melanoma drug resistance. Roth et al (85) developed pooled knockin sequencing (PoKI-seq), a technology that measures cell abundance and state both ex vivo and in vivo. This method facilitates the barcoding and tracking of targeted integrations of large non-viral DNA templates in primary human T cells. The technology notably identified a novel TGF-β R2-41BB chimeric receptor, enhancing the clearance of solid tumors. PoKI-seq enables the parallelized rewriting of endogenous genetic sequences, accelerating the identification of effective knockin programs for cell therapies. However, specific applications of single-cell CRISPR technology in GC research have yet to be discovered.

Notable advancements have been made in integrating single cell multiple-omics analyses. For instance, Trio-seq allows for the simultaneous analysis of the genome sequence, epigenome and transcriptome in a single cell (86). Another technique, cellular indexing of transcriptomes and epitopes by sequencing, combines surface protein analysis with transcriptome sequencing (87), while Single Cell Methylome and Transcriptome Sequencing enables the simultaneous analysis of both the epigenome (specifically DNA methylation patterns) and transcriptome (gene expression profiles) at the single-cell level, and combines DNA methylation analysis with transcriptome sequencing within a single cell (88).

In conclusion, single-cell omics technologies, encompassing SCS, single-cell proteomics and multi-omics have undergone advancements, enabling researchers to explore the intricate details of cellular heterogeneity across various omics layers. These technologies offer insights into gene expression, genetic variations, epigenomic modifications and protein expression patterns at the single-cell level, enhancing the understanding of cellular dynamics and disease mechanisms.

Early applications of SCS in GC focused on transcriptome and single-cell genome analysis of GC cell lines and reported marked genetic and transcriptional diversity (93). In one study, the identification of 24 notable mutated genes among tumor cells demonstrated the genetic alterations underlying GC and potential therapeutic targets (94). Analysis of circulating tumor cells (CTCs) from patients with advanced GC demonstrated numerous mutations in the genes associated with the KRAS and Rap1 pathways, as well as mutations in the genes associated with the MET/PI3K/AKT pathway and the SMARCB1 gene in patients with large multiploid CTCs (95), leading to the development of resistance to either chemotherapy alone (96) or chemotherapy combined with targeted therapy (97) in patients with GC.

In a study on GC lymph node metastasis, scRNA-seq was performed on primary and metastatic tissues from 3 patients, revealing intratumoural heterogeneity and distinct carcinoma profiles. The results identified a subgroup of cells indicating a transitional state in the metastasis process, and also revealed potential marker genes (ERBB2, CLDN11 and CDK12) and genes driving gastric cancer evolution (FOS and JUN), offering insights for GC treatment (98). A panel of biomarkers was identified for discriminating between benign and malignant epithelial tissues, potentially aiding in early detection and diagnosis of GC (99). Using scRNA-seq analysis, subtypes of peritoneal carcinomatosis samples from patients with GC were classified, and a 12-gene prognostic signature was identified (100). Investigation of metastatic GC using peritoneal ascite samples and cerebrospinal fluid revealed that poor prognosis was associated with M2-like characteristics in tumor-associated macrophages (101). In a study of patients with non-metastatic GC, immunosuppressive gene expression patterns were enriched in regulatory T cells (Tregs) within gastric tumor tissues, indicating an immunosuppressive TME. The absence of a separate exhausted CD8⁺ T cell cluster and low expression levels of exhaustion markers were also observed, and ACKR1 was identified as a potential marker associated with poor prognosis (102). Using scRNA-seq, a broad spectrum of GC subtypes was assessed to create a transcriptomic map of biomarkers from malignant epithelial cells for the prediction of overall survival in patients with GC (103). OR51E1 has been identified as a key marker gene for unique endocrine cells in early-malignant lesions of gastric cancer, offering a potential avenue for early detection of malignancy (103). Simultaneously, HES6 has been recognized for its potential utility in identifying metaplasia at an early stage, demonstrating its importance in the early diagnosis and intervention of precancerous gastric conditions (104). Furthermore, a panel of early GC-specific signatures was identified using mucosa biopsies, which may be used in clinical applications for early diagnosis (104).

Analysis of the TME in patients with GC revealed increased stromal cells and Tregs, unique transcriptional cell states in dendritic cells (DCs), exhausted cytotoxic T lymphocytes and a specific extracellular matrix composition not found in normal tissue (105). Furthermore, in more advanced disease stages, a downregulation of interferon regulatory factor 8 in CD8⁺ tumor-infiltrating lymphocytes has been reported, revealing changes in the immunological landscape in GC and its potential implications for disease progression (106).

Another study demonstrated that monocyte-like DCs and autophagy-related genes marking high-plasticity (ability of certain cancer cells to adapt and change in response to different environments or therapeutic pressures) GC were associated with poor prognosis during GC peritoneal metastasis progression (107). In tumors with a high alternate promoter burden (APB-high), characterized by increased use of alternative gene promoters, distinct immunological populations were observed along with a reduced proportion of T cells. These findings shed light on the immunological aspects of GC and how the APB-high status influences tumor progression and the immune response (108).

Comparisons between GC and colorectal cancer revealed distinct mutational landscapes and microbiomes, contributing to differences in the TME, and thus, disease prognosis (109). Furthermore, the communication between cancer-associated fibroblasts (CAFs) within the TME and other cells provides insights into their regulatory functions (110).

Activated fibroblasts, endothelial cells, immunosuppressive myeloid cell subsets and Tregs present in the TME were associated with an unfavorable prognosis and resistance to anti-programmed cell death 1 therapy in patients GC (91). Characteristics linked to a positive response to platinum-based chemotherapy were defined, aiding personalized treatment decisions. For example, response was associated with on-treatment TME remodeling, including natural killer cell recruitment, decreased tumor-associated macrophages, M1-macrophage repolarization and increased effector T-cell infiltration (91). In non-responders to chemotherapy, Kim et al (111) observed low or no programmed death-ligand 1 expression, an increase in Wnt signaling and B-cell infiltration, a higher presence of lymphocyte activating 3-expressing T cells, and a reduction in dendritic cells. This suggests a distinct pattern of immune changes associated with chemotherapy resistance.

To assess the effects of combination therapy with camrelizumab and 5-fluorouracil, leucovorin and oxaliplatin on GC and its impact on the TME, Li et al (112) conducted single-cell RNA sequencing on 10 GC samples both before and after neoadjuvant treatment. This study highlighted that high expression of interferon-γ in CD8⁺ T cells was associated with enhanced responses to this combination therapy, indicating an immunological impact on the tumor environment (112). Additionally, a murine model suggested the potential therapeutic approach of combining anti-IL-17 and anti-programmed death-1 monoclonal antibodies for GC tumor regression (113).

In conclusion, these studies (112,113) demonstrated the wide-ranging application and use of single-cell omics technologies in GC research. By analyzing single cells, these studies have provided insights into intratumoral heterogeneity, the TME, immune and treatment responses, prognostic markers, and potential therapeutic targets. In addition, these techniques contribute to the understanding of GC biology and hold promise for improved diagnostics and personalized treatments.