Forging new insights in forensic microbiome studies: A 2000 to 2024 bibliometric analysis redefining the landscape

Abstract

Background:

Microbial diversity has been extensively studied across various fields, including medicine and therapeutics. Its application in forensics is rapidly expanding due to its effectiveness. This study aimed to conduct a comprehensive bibliometric analysis of global research on forensic microbiome, providing a foundational knowledge framework for this emerging field.

Methods:

A comprehensive literature search was performed using the Web of Science Core Collection database to identify publications related to the forensic microbiome. Annual publication output, research collaborations, research hotspots, and developmental trends in this field were analyzed using bibliometric software (VOSviewer and CiteSpace).

Results:

In total, 709 articles published between 2000 and 2024 were selected. The first study in this field was published in 2000. Recently, the number of publications and citations has grown significantly. Cooperation network analysis revealed that the United States of America contributes the most to forensic microbiome research, with the highest publication volume. Michigan State University emerged as the most prolific institution. Forensic Science International is the most productive journal in this field. Carter David O. contributed the most to the articles and is the most co-cited authors. Keywords cluster analysis identified 4 major research clusters: bacteria, forensic medicine, 16s ribosomal ribonucleic acid, and machine learning. Machine learning, human microbiome, and forensic microbiology have attracted increasing attention from researchers.

Conclusion:

This bibliometric analysis provides a data-driven and objective overview of forensic microbiome research currently, offering readers a valuable reference for future research. Our review provides insights into contemporary trends, global patterns of collaboration, fundamental knowledge, high-interest research areas, and emerging frontiers in the forensic microbiome.

1. Introduction

Microbiomes (e.g., viruses, bacteria, and fungi) are ubiquitous, forming remarkable diverse communities with certain physical and chemical properties (including microorganisms and their environment).^[1] Microbial communities are influenced by geographic, ethnic, lifestyle, and environmental factors, such as urbanization.^[2]

The forensic microbiome refers to the study of the microbial communities associated with forensic evidence, such as those found on human remains, in soil, on personal items, or at crime scenes. It involves analyzing the composition, diversity, and dynamics of these microbial populations to gain insights that can aid criminal investigations.^[3] Forensic microbiome is a relatively new field in forensic science, encompassing the application of scientific methods to analyze evidence related to bioterrorism, biocrimes, hoaxes, or the accidental release of a biological agent or toxin for attribution purposes.^[4] The microbiome profile of a soil sample can be used to establish its geographic origin, associate humans and animals with a specific area, or estimate the postmortem interval.^[3] The human microbiome can help identify the type of stains in saliva and vaginal fluid, as well as attribute the stains to individuals. With advanced methodologies for the detection, identification, and characterization of microorganisms, significant progress has been made in forensic microbiome.^[5]

Forensic microbiome research has witnessed remarkable growth in the number of published studies, as researchers from diverse backgrounds, including microbiology, forensic science, and bioinformatics, have collaborated to explore its possibilities.^[1,3] However, despite this progress, a comprehensive understanding of the overall development and current state of forensic microbiome research remains incomplete. This is where bibliometric analysis plays a key role. By systematically examining the body of literature in this domain, we can uncover hidden patterns, trends, and relationships that might otherwise go unnoticed.

Bibliometrics is a literature analysis method that quantitatively and qualitatively examines the output and status of publications.^[6] It can be used to analyze published information and related metadata, employing statistical methods to describe or illustrate relationships between scholarly works.^[7] Commonly used bibliometric tools include CiteSpace (version 6.3. R1; Drexel University, Philadelphia), VoSviewer (version 1.6.20; Leiden University, The Netherlands), and R package bibliometrix.^[6,8,9] In 2023, Guo et al summarized the global research status of microbial forensics.^[10] While their work has provided valuable insights, it was limited in scope and did not report the detailed research progress of forensic microbiomes. To address this knowledge gap, we conducted a more in-depth analysis extending beyond Guo et al (2023). Specifically, this study aimed to perform a detailed bibliometric analysis of publications on forensic microbiome over the past 2 decades (from 2000–2024) to identify major contributors and current research status, emphasizing research trends and future developments in this field.

2. Materials and methods

2.1. Data collection

This study focuses on the field of forensic microbiology, aiming to conduct a systematic bibliometric analysis of relevant studies in this interdisciplinary field. Web of Science is one of the largest databases frequently used to retrieve relevant literature in different scientific fields. We conducted a literature search of the web of science core collection (WoSCC) (https://www.webofscience.com/wos/woscc/basic-search) by the same investigator on a single day (January 26, 2025). The publication period for this study was set from 2000 to 2024. The search terms were: Topic = “microbial” OR “microbiome” OR “microbiology” AND Topic = “forensic.” And the terms were searched in the title, abstract, and keywords fields of the database. After the initial search, publications were screened based on the following inclusion criteria: the publication time span was set from January 1, 2000, to December 30, 2024; only English-language publications were included; and categorized as original articles or reviews. All documents related to forensic microbiome were exported in “full records and references” text file format. VOSviewer and CiteSpace were used to retrieve and import for bibliometric analyses.^[5,10] Exclusion criteria were specified, and the retrieved documents were deduplicated. Two independent reviewers conducted a preliminary screening of the retrieved studies according to preestablished inclusion and exclusion criteria. Inclusion criteria focused on research topics that were directly related to the forensic microbiome. If there is a disagreement between the 2 people, a consensus will be reached through discussion, reference to third-party opinions, or backtracking to the original literature according to the detailed resolution process. To enhance the reliability and accuracy of the bibliometric analysis, 2 independent reviewers manually validated the keyword clusters and thematic classifications derived from the automated software. Any discrepancies were resolved through discussion until a consensus was reached. A comprehensive workflow design for literature screening and data analysis is presented in Figure S1.

This study is of a review nature and does not involve any human or animal experiments. All data are sourced from published literature. Therefore, there is no need for approval from the ethics committee.

2.2. Data analysis and visualization

VOSviewer (version 1.6.20; Leiden University) is bibliometric analysis software that visualizes complex data relationships via intuitive network diagrams. It was employed to conduct network visualization analysis of the countries, institutions, journals, and keywords with relevant publications. In the maps generated by VOSviewer, each node represented an item, such as a country, institution, journal, or author. The size and color of the nodes indicated the frequency and classification of these items, whole the thickness of lines between nodes reflected the strength of collaboration between them.^[9]

CiteSpace software (version 6.3. R1; Drexel University) supports data conversion from multiple databases and includes built-in deduplication tools. It also offers time-slicing processing and diverse screening criteria, which effectively clean and preprocess data to enhance the accuracy and reliability of the analysis results. It was used to generate keyword-burst intensity maps. Keyword emergence referred to the sudden and frequent use of keywords within a certain period to reflect hot topics and future trends in a research field.^[11]

CiteSpace exhibits stronger data preprocessing and screening capabilities, enabling it to handle various data formats and noise issues. VOSviewer holds unique advantages in data visualization, presenting data in a more beautiful and intuitive manner. In practical applications, the data preprocessed by CiteSpace can be imported into VOSviewer for visualization to obtain better-looking and more informative visual results.^[9,12]

3. Results

3.1. Analysis of publication trends

In total, 954 relevant research documents were retrieved, and 709 pieces of literature were included in the final analysis. Of these, 555 were original research articles and 154 were review articles, respectively. The first of such study was published in 2000. Table 1 presents the annual distribution of these articles. Few articles were published during the initial phase (2000–2008). Publication volume grew steadily from 2009 to 2020 and a marked surge in research activity from 2021 to 2024, reflecting increased interest in this field.

Table 1.

Annual distribution of the number of papers (2000–2024).

Year	Papers	% of 709
2024	86	12.13
2023	57	8.04
2022	59	8.32
2021	63	8.89
2020	50	7.05
2019	40	5.64
2018	37	5.22
2017	47	6.62
2016	45	6.35
2015	35	4.94
2014	22	3.10
2013	25	3.53
2012	12	1.69
2011	16	2.26
2010	18	2.54
2009	19	2.68
2008	9	1.27
2007	15	2.12
2006	7	0.99
2005	18	2.54
2004	9	1.27
2003	7	0.99
2002	6	0.85
2001	3	0.42
2000	4	0.56

3.2. Distribution of countries and institutions

3.2.1. Contributions of countries

We used VOSviewer software to analyze the data, generating national visualization maps (Fig. 1). The results revealed that the top 5 countries by publication count were the United States of America (USA) (279, 39.35%), China (90, 12.69%), England (60, 8.46%), Australia (52, 7.33%), and Italy (51, 7.19%) (Table 2, Fig. 1). These findings indicate that the USA holds the leading position in forensic microbiome research.

Figure 1.

Network visualization map of co-citations of countries. The larger the circle node in the figure, the more papers published. The node connection represents the association strength; the thicker the connection, the more time the 2 countries cooperate in publishing papers. Node colors represent different clusters.

Table 2.

The top 10 countries with the most reviews and total citations.

Number	Countries	Document	Citations
1	USA	279	9909
2	China	90	1060
3	England	60	2080
4	Australia	52	2418
5	Italy	51	1026
6	Spain	32	1164
7	Germany	26	715
8	Nertherlands	25	872
9	India	24	183
10	Canada	20	1043

USA = United States of America.

3.2.2. Contributions of institutions

The field of forensic microbiome has attracted attention of 1085 academic institutions. Figure 2A displays an institutional visualization map generated with VOSviewer. Michigan State University emerged as the leading contributor with 25 publications. Others included Southern Medical University, University of Tennessee, Chaminade University of Honolulu, and Central South University (Table 3). Institutional analysis revealed that the majority of these institutions were from the USA (Fig. 2A). Figure 2B lists the top 10 cited institutions ranked by the duration of their citation prominence, highlighting the significant impact of these institutions.

Figure 2.

Visualization of institutions. (A) Institutional co-citation network (Node size: institutional publication count, edge thickness: inter-institutional co-citation strength, colors: research clusters); (B) top 10 institutions with the strongest citation bursts, highlighting rapidly emerging institutional research impact.

Table 3.

The top 15 institutions with the most reviews and total citations.

Number	Institutions	Document	Citations
1	Michigan State University	25	669
2	Southern Medical University	18	182
3	University of Tennessee	17	685
4	Chaminade University of Honolulu	15	878
5	Cent South University	12	145
6	Western Australia University	11	902
7	Calif San Diego University	11	697
8	Alabama State University	11	467
9	Pacific NW National Lab	11	205
10	Virginia Commonwealth University	11	196
11	Minist Public Security	11	43
12	Colorado University	10	1319
13	Texas A&M University	10	620
14	Adelaide University	10	323
15	Netherlands Forens Institute	10	312

3.3. Analyses of authors

Examining the seminal works of impactful authors in a field provides insights into its foundational theories. In total, 2984 authors contributed to 709 publications, yielding a co-authorship index of 4.21. Table 4 lists authors with 10 or more publications. The results showed that Carter David O. (Chaminade University of Honolulu, USA), Chen Ling (Southern Medical University, China), and Budowle Bruce (University of North Texas Health Science Center, USA) had the most prolific output and citations. The author co-citation visualization map and top ten cited authors (Fig. 3) demonstrated that these authors were densely cross-cited and exerted significant influence.

Table 4.

The top 10 authors with the most reviews and total citations.

Number	Authors	Document	Citations
1	Carter, David O.	20	1697
2	Chen, Ling	14	83
3	Budowle, Bruce	13	421
4	Benbow, M. Eric	13	360
5	Javan, Gulnaz T.	11	467
6	Pechal, Jennifer I.	11	298
7	Procopio, Noemi	11	145
8	Liu, Chao	11	80
9	Finley, Cheree J.	10	461
10	Devruyn, Jennifer M.	10	436

Figure 3.

Visualization of authors. (A) Network visualization map of co-citations of authors (Node size: author publication output, edge thickness: inter-author collaborative relevance, colors: author clusters); (B) top 10 authors with the strongest citation bursts, identifying authors with surging academic influence.

3.4. Analyses of journals

The field of forensic microbiome attracted contributions from 222 journals. Table 5 lists the top 15 journals with the most publications in this field. Frontiers Science International was the most prolific journal, with 73 articles, followed by Journal of Forensic Sciences (69), Forensic Science International-Genetics (44), Forensic in Microbiology (36), and International Journal of Legal Medicine (32). Journal of Forensic Sciences had the highest citation rate (1379 citations). These results indicate the high quality of the articles published in these journals and their high scientific value and impact.

Table 5.

The top 15 journals with the most reviews and total citations.

Number	Journals	Document	Citations
1	Frontiers Science International	73	1229
2	Journal of Forensic Sciences	69	1379
3	Forensic Science International-Genetics	44	1083
4	Forensic in Microbiology	36	528
5	International Journal of Legal Medicine	32	561
6	Journal of Forensic and Legal Medicine	26	292
7	Plos One	17	643
8	Applied and Environmental Microbiology	13	410
9	Scientific Reports	12	344
10	Science and Justice	12	194
11	Forensic Science Medicine and Pathology	10	136
12	Enviromental Forensics	10	96
13	Microorganisms	10	94
14	Msystems	7	211
15	Legal Medicine	7	83

3.5. Key topics of research hotspots

The core topics addressed in scholarly papers are reflected in keywords and high-frequency keywords, which are suitable for inclusion in co-occurrence analysis.

3.5.1. Analysis of clusters and co-occurrence of keywords

VOSviewer was used to construct keyword co-occurrence network visualization for the 709 articles. A total of 3538 keywords were retrieved for this study. Based on the link strength of keyword co-occurrence, we selected 38 keywords with a frequency of occurrence ≥ 20 times for visualization and divided the network into 4 clusters. Statistical analysis of keywords across different sections of the paper identified distinct research direction clusters. Cluster analysis of these keywords elucidated the knowledge structure of the field. Clusters 1 (red) and 2 (green) were the largest, each containing 11 terms. The core themes in Cluster 1 (red) were microbiome, diversity, identification, and forensic microbiome, whereas Cluster 2 (green) was primarily associated with bacteria, forensic science, microbiology, diagnosis, and samples. Cluster 3 (blue) focused on DNA, soil, extraction, and bacteria. Cluster 4 (yellow) was related to decomposition, potential use, time, and succession (Figs. 4A and 4B).

Figure 4.

Visualization of keywords. (A) Network visualization map of co-citations of keywords (Node size: keyword occurrence frequency, edge thickness: keyword co-occurrence strength, colors: thematic clusters); (B) analysis results of hotspot and topic migration in the field of forensic microbiome; (C) keyword clustering timeline, revealing temporal dynamics of research themes; (D) top 10 keywords with the strongest citation bursts, marking emerging field focal topics.

3.5.2. Burst detection and overlay visualization of keywords

Keyword-burst detection enables rapid identification of emerging research hotspots in a specific field. Figures 4C and 4D display the most robust keyword bursts in the forensic microbiome research from 2000 to 2024. “Microbial forensics” was the most robust burst keyword (strength 10.04) from 2004 to 2015, followed by “forensic science” (strength 8.23, 2006–2013), “forensic microbiology” (strength 6.23, 2019–2024), “body fluid identification” (strength 6.08, 2022–2024), and “death” (strength 5.31, 2022–2024), revealing that these keywords represented the popular research topics in recent years and even in the near future.

4. Discussion

4.1. General information

We used VOSviewer and CiteSpace to visualize the spatial and temporal distributions of the literature, keywords, publication citations, and co-citations, providing a comprehensive overview of forensic microbiome research.

In this study, 709 articles, published between 2000 and 2024, were included. The first study was published in 2000. These indicate increment in the annual number of publications. From 2000 to 2015, research in this field remained in its infancy, with an average 14.06 papers per year. From 2016 to 2024, publications increased significantly, averaging 60.50 papers annually. The number of related publications has grown rapidly over the past 8 years, indicating that forensic microbiome research is on the rise. In this field, the top ten countries accounted for 92.94% of the total publications. The USA and China exhibit the deepest academic accumulation and strongest influence. The USA demonstrated the highest level of international collaboration and had the highest publication counts. Our results revealed that the USA played a leading role, with 9909 citations across 279 articles (average 35.52 citations per article). The USA also hosts the top ten productive research organizations, with Michigan State University being the most productive institution worldwide, which can be attributed to their top researchers and well-established forensic microbiome foundations. These findings highlight the significant contributions and potential scientific innovations of USA efforts in forensic microbiome research. Furthermore our co-cited author network and author’s publication volume analysis underscore David O Carter (Chaminade University of Honolulu) to the field.

Although cooperative relationships exist among some countries, the breadth and intensity of cooperation between institutions remain suboptimal. For instance, limited cooperation is observed between institutions in the USA and China, while stronger collaborative ties exist among countries and institutions in Europe and USA. Research competitiveness can be enhanced through international teamwork, highlighting the importance of extensive collaboration between institutions. It is crucial to motivate and support more research efforts from less advantaged countries or institutions. This will enable a more comprehensive understanding of how diverse sociocultural elements interact within the context of forensic microbiology.

Approximately 94.84% of the affiliations had 5 articles, indicating that most affiliations have not made in-depth investments and only a few affiliations have conducted continuous research. Among journals interested in this topic, Frontier Science International is the most productive.

Collectively, these bibliometric findings reveal that forensic microbiome has evolved from a niche theoretical topic to a rapidly expanding interdisciplinary field, driven by technological advances and growing forensic application demands. The uneven distribution of publications, citations and collaborative networks underscores the urgent need for more inclusive global partnerships to bridge resource gaps and enhance the representativeness of research data across diverse geographic and socioeconomic contexts. Moreover, the fragmented institutional output and lack of specialized journal platforms suggest that sustained investment, standardized research protocols, and targeted academic exchanges are critical to maturing this field and translating basic findings into practical forensic tools.

4.2. Research hotspots and frontier exploration in forensic microbiome

The keywords summarized themes in the literature. The frequency with which keywords appear in the literature can reveal research hotspots in the field. Keyword co-occurrence analysis can indicate the study category and research hotspots in the field, and display discovery trends. “Microbial forensics” was the most robust burst keyword (strength 10.04) from 2004 to 2015, followed by “forensic science” (strength 8.23, 2006–2013), “forensic microbiology” (strength 6.23, 2019–2024), “body fluid identification” (strength 6.08, 2022–2024), and “death” (strength 5.31, 2022–2024), revealing that these keywords represented the popular research topics in recent years and probably in the near future. Machine learning, human microbiome, and forensic microbiology are relatively recent keywords that have attracted increasing attention from researchers. Such features can help scholars better integrate the study of microorganisms into the practical applications of forensic medicine. The rising prominence of machine learning and human microbiome keywords signifies a critical interdisciplinary integration trend, which holds great promise for optimizing microbial biomarker screening and improving the accuracy and efficiency of forensic detection technologies in subsequent investigations.

4.3. Main focus of research in forensic microbiome

Forensic science uses physical evidence to solve investigations based scientific principles, thus operating within a theoretical framework.^[13] Numerous microbiome studies have demonstrated that the application of microorganisms can assist forensics in areas such as trace evidence, source tracking, and circumstances of death. However, most forensic microbiome studies have focused on soil and human samples.^[14]

Soil is rich in bacteria, fungi, viruses, and other microorganisms, and its polymorphic composition significantly affects its properties and environment. Polymorphisms in soil microorganisms have regional, temporal, and environmental impacts.^[15] Soil evidence is a powerful tool in criminal cases. The ubiquity and transferability of soil make it an important resource for forensic investigators, as it can establish a link between agents and scenes.^[15,16] The ability of soil to transfer to items such as shoes and clothing and its persistence can vary depending on the soil characteristics, mineralogy, and type of contact. It can be used to find clues about the origin of an unknown specimen or link between a crime scene and a suspect.^[15–17] Microbial DNA in soil is relatively easy to access and analyze, providing a powerful means of distinguishing soil samples or linking them to a common origin.^[15] Areas of exploration included the individual identification of microorganisms, the relationship between changes in soil bacteria, and the time of death.

The numerous microbial communities harbored by the human body are collectively known as the human microbiome.^[18] Human-associated microbiomes can be shed, deposited, and transferred in a process similar to that of human cells. When these microbial communities are left at a crime scene, they can be detected by forensic scientists and provide useful information.^[4] Specific microbial species obtained from crime scenes provide evidence of human contact with body fluids.^[2] Human microbiomes can be collected from sources such as skin, body fluids, houses, and personal belongings.^[19–21] The human microbiome can be applied for forensic investigations, such as assessing individual variability via skin microbiomes, reconstructing crime scenes, and linking evidence to specific individuals.^[2,22]

Genotyping involves analyzing the genetic material, typically DNA, of microorganisms present in various forensic samples. Each microorganism has a unique genetic sequence, by examining specific regions of the DNA, scientists can identify and distinguish different microbial species and strains.^[23] Genotyping has a broad range of applications, including genealogy, assessing risks and causes for common diseases and health conditions, and forensic investigations.^[24] Probabilistic genotyping and its associated software has greatly aided in forensic DNA mixture analysis.^[25] Geolocation connects an individual to a certain place or location by microbial data.^[1] DNA recovered from dust and environmental debris can be used to identify the organisms associated with these sample types. Then be leveraged to discern sample origin or geolocation and investigate individual identification.^[26]

The application of microbial community diversity analysis in forensic science remains in its infancy. However, molecular biology tools, such as DNA fingerprinting, whole-genome sequencing, and microarray analysis, have significantly advanced the field of microbial forensics over the last 2 decades.^[27] Artificial intelligence (AI) is increasingly being applied to microbial data processing owing to its powerful recognition, processing, and data analysis capabilities, with ML being used as a subset of AI algorithms.^[28,29] Several studies have been conducted in areas including cause-of-death analysis, postmortem interval inference, postmortem submersion interval inference, forensic individual identification, tissue or body fluid source identification, and geographic location tracing.^[30,31] MI algorithms can handle vast amounts of complex microbiome data that would be overwhelming for traditional analytical methods. MI models can be used as black boxes to predict a specific outcome and a powerful tool that can provide deep insights into microbial communities and identify patterns in microbial community data.^[32] Deep learning models, including the convolutional neural network model and convolutional neural network-derived models, enhance the accuracy of forensic prognosis.^[33]

4.4. Future development

Numerous studies have been conducted on forensic microbiome; however, it is still at a superficial level, and it is believed that with the persistent efforts of researchers, forensic microbiome will eventually provide reliable evidence for judicial cases and promote the progress of forensic medicine. Considering the increasing availability of high-throughput sequencing data and advancements in computational power, future studies could focus on developing more accurate and rapid microbiome profiling methods. To further promote clinical translation and practical application, standardized experimental protocols and unified analytical pipelines are urgently needed to improve the reproducibility and comparability of forensic microbiome research. In addition, interdisciplinary cooperation combining multi-omics strategies, AI, and international collaborative networks will greatly accelerate the development and practical application of microbiome-based evidence in real forensic cases. And interdisciplinary studies that combine forensic microbiome analysis with other emerging fields like AI to solve complex forensic problems would be a great addition.

4.5. Limitations

This study has some limitations. The data for this study were only obtained from the WoSCC database; other databases were ignored, which may have omitted some relevant studies. Although WoSCC is widely recognized as a comprehensive and authoritative database covering high-impact journals worldwide, relying on a single database may lead to the omission of relevant publications indexed primarily in other platforms, potentially resulting in an incomplete overview of the research field. We filtered studies published in English, underestimating non-English-written papers. This restriction may lead to the underestimation of research contributions from non-English-speaking countries and regions, limiting the global representativeness of the analysis. These limitations were mainly due to considerations of database accessibility, standardized indexing, and analytical feasibility. Despite these constraints, the findings still provide valuable insights into the current status and trends of the field. Future research can adopt a more inclusive strategy by incorporating multiple databases and multilingual literature to achieve a more thorough and balanced review.

5. Conclusion

In conclusion, we systematically analyzed the literature on forensic microbiomes using a bibliometric approach. Compared with traditional reviews, this review offers original and objective insights for research on forensic microbiomes and serves as a valuable reference for further studies. Microbiomes hold significant research value and promising application prospects in forensic medicine. The rapidly growing number of publications indicates that forensic microbiome research is gaining increasing recognition among scholars worldwide. The leading countries are the USA and China; however, inter-country and inter-institutional cooperation and communication remain to be strengthened.

Acknowledgements

We would like to thank Editage (www.editage.cn) for the English-language editing.

Author contributions

Conceptualization: Zheng-Jiang Wang, De-Xia Wang, Chun-Yan Wang, Mai-Qing Yang.

Data curation: Zheng-Jiang Wang, De-Xia Wang, Chun-Yan Wang, Mai-Qing Yang.

Formal analysis: Zheng-Jiang Wang, De-Xia Wang, Chun-Yan Wang.

Funding acquisition: Gui-Yan Han.

Methodology: Zheng-Jiang Wang, Mai-Qing Yang.

Supervision: Gui-Yan Han, Mai-Qing Yang.

Writing – original draft: Gui-Yan Han, Zheng-Jiang Wang.

Writing – review & editing: Gui-Yan Han, Mai-Qing Yang.