The journey of F1000Research since inception: through bibliometric analysis

Study design

In 1969, Pritchard introduced Bibliometrics as “applying mathematical and statistical methods to books and other means of communication” (Pritchard, 1969). In the present scenario, it is widely seen in every field. It is a quantitative and qualitative research analysis used to understand and highlight the impact of authors, institutions, collaborations, emerging research areas and countries. The present study used a bibliometric technique through VOS-viewer and Biblioshiny (R-studio) to analyse the performance of the F1000Research since its inception (2012). It is a Q1 category journal indexed in Scopus, UGC CARE, DOAJ, and PubMed. Scopus is one of the best reliable databases, which has 1.8+ billion cited references, 84+ million records, 17.6+ million authors’ profiles, 94.8+ affiliation profiles, and 7+ thousand publishers, where 35% of publications are from the field of social sciences, 27% physical sciences, 23% health sciences and 15% life sciences (ELSEVIER, 2020). Hence, the Scopus database has been used to extract the metadata, as suggested by Baber et al. (2023).

Metadata extracted from the Scopus database has been analysed using VOS-viewer software developed by van Eck & Waltman (2010) and Biblioshiny (R-studio) interface developed by Aria & Cuccurullo (2017). Bibliometric metadata represents various relationships, such as publications and citations (Ding et al., 2017). Bibliometric analysis helps connect the journal’s several variables, which future researchers can use to move towards the proper direction (Kaurav & Gupta, 2022). The impact of the publications is measured through their citations, whereas the number of publications only quantifies their productivity (Svensson, 2010). H-index explains “the number of papers with citation number >h, as a useful index to characterise a researcher’s scientific output” (Hirsch, 2005), which has also been used in the present study.

Data collection

Various studies have used keywords search index criteria (Chen et al., 2017; Maier et al., 2020; Kumar et al., 2022; Leong et al., 2021; Pesta et al., 2018; Singh et al., 2022) for extracting the Scopus metadata. It has prepared the base and design for data collection, which has also been used in the present study.

Step 1: The Scopus database was searched with the keyword (ALL(F1000Research) AND PUBYEAR > 2011 AND PUBYEAR < 2023 AND (LIMIT-TO (EXACTSRCTITLE,“F1000research”)) AND (LIMIT-TO (LANGUAGE,“English”)) AND (LIMIT-TO (DOCTYPE,“ar”) OR LIMIT-TO (DOCTYPE,“re”))) in the ‘source title’ on 22^nd March 2023 from the IP address of ‘Manipal Academy of Higher Education, Manipal, India.’ Metadata of 72396 documents appeared in the initial search.

Step 2: Various journals appeared but were limited to an ‘F1000Research’ only., in which 5999 metadata appeared.

Step 3: The result from 2012-2022 (31^st December 2022), generated 5987 documents and then refined to the language ‘English,’ publication stage ‘Final’, limited to ‘article and review’ only generated 5767 documents.

Step 4: Only 5727 metadata were downloaded, whereas the remaining 40 could not be downloaded due to a metadata error. Finally, a CSV file, including “Citations information, bibliographical information, abstracts and keywords, funding details and other information,” was downloaded.

Step 5: Purification of data performed in CSV file using the conditional formatting features to remove the duplicate articles (202 articles) and removed seven articles due to retraction case. Finally, 5518 articles (metadata) were used for the final analysis using VOS-viewer software and Bibliometrix (R-studio) interface (Kumar, 2023).

Data analysis

The present study used VOS-viewer software version 1.6.19 and Biblioshiny (R-studio) 4.2.3 interface to analyse the F1000 research journey. A samples of 1000 papers considered as an good enough to generalise the result (Rogers et al., 2020). To analyse the publication trends of F1000Research, find out the leading authors, organisations, countries, frequent keywords, most cited publications, bibliographic coupling between authors, organisations and countries, and emerging research areas since its inception VOS-viewer and Biblioshiny were used. VOS-viewer uses a fractional counting method, whereas the Scopus database generates metadata under a full counting system; therefore, full and fractional both counting methods have been used. However, Biblioshiny (R-studio) interface generates a detailed analysis. However, data reading showed poor keywords details, a completely missing number of cited references, and a completely missing science categories, which specifies some missing data. Although extracted metadata provided enough information to be used in the final analysis.

Publication trends and citation analysis of F1000Research

Quantifying publications and their citations play a significant role in assessing the journal’s global presence and recognition. In the present study, Figure 1 represents the number of publications that remained almost constant during 2016, 2017, 2020, and 2021 (ranging between 614-669), whereas, in 2018, publications increased drastically to 915 and again reduced in 2019 to 764. It shows that before COVID-19, upward growth in the publication was seen due to the journal’s recognition and popularity among the researchers. Citations which make the article journal more impactful, increased dramatically in 2016 to 16301 from 5162 (2015) and again started declining. Citations formed a bell-curve shape, which shows that they grew from 413 citations in 2012 to 16301 and reduced to 276 in 2022. New articles are still to create their impact in the researcher’s mind to cite them in their literature. F1000Research started its journey in 2012 and published only 43 articles, multiplied thrice in consecutive years.

Figure 1. Publications and Citations trends since 2012 (Source: Scopus Database, Software: MS-excel).

Organisations contributions (Documents & Citations)

Table 2 represents the top 20 organisations that contributed to the F1000Research journal. A maximum number of articles published by “Faculty of Management, Multimedia University, Cyberjaya, Selangor, Malaysia,” i.e., 12 but got only seven citations as they have published their work recently (2021). Whereas the maximum number of citations received by “SIB Swiss Institute of Bioinformatics, University of Zurich, Zurich, Switzerland,” i.e., 1834, is exhibited in Figure 2. However, it has published only four articles. Following the “Department of Microbiology, School of Medicine, Universitas Syiah Kuala, Banda Aceh, Aceh, Indonesia” published seven articles having a citation of 211. “Brawijaya Internal Medicine Research Center, Department of Internal Medicine, Faculty of Medicine, Universitas Brawijaya, Malang, East Java, Indonesia”, ranked number three, has published eight documents but has g-citations of 202.

Figure 2. Overlay graphical representations of the top organisation (Data source: Scopus, Software: VOS-viewer).

European, Asian (mainly Malaysia, Indonesia, and Saudi Arabia), and American (both south and north) organisations contributed to the top 20 organisations list. In contrast, African and Australian organisations are not in the top 20 organisations listed in Table 2. The organisation that published their articles in the F1000Research journal from 2016-2022 (22^nd March) is represented in Figure 2. Organisations marked in yellow (“Faculty of Computing and Informatics,” “Faculty of Management,” “Faculty of Engineering,” “Molecular Diagnostic Laboratory,” and “Department of Community Medicine”) are relatively young in publishing their work in the F1000Research journal.

Country’s production overtime

Table 3 represents the country’s production over time, citations generated by the specific country and total link strength (TLS). Figure 3 illustrates the significant countries’ contributions/production over time (2012-2022). According to Worldometer (2023), there are 195 countries in the world, of which 164 countries have contributed their work to the F1000Research journal, ranging from 1 to 1893 publications. A major contribution was from the United States, United Kingdom, and Germany, respectively, i.e., 1893, 882 and 357. However, citations from the United States, United Kingdom, and Canada ranked in the top three (31765, 12866 and 5776, respectively) as shown in Table 3. India has also contributed 274 publications (7^th in publication ranking), citations of 1589 (13^th in citations ranking) and a TLS of 17025. A major chunk of Africa, Central Asia, and Greenland have contributed their work to the F1000Research journal, as shown in Figure 3. For global recognition and contribution, the journal should target the countries whose contribution is either less or not contributed. Quality publications can come from even the world’s smallest countries; therefore, more focus must be given to those countries.

Figure 3. Graphical representation of countries’ contribution to the F1000Research journal (Data source: Scopus, Software: MS-excel).

Most frequent keywords and their occurrence

The most frequent keyword visualisation shown in Figure 4 has been generated using the Biblioshiny (R-studio) interface. These keywords are based on the co-occurrence of author keywords, which helps understand the most impactful keywords and keyword popularity. Table 4 exhibits the top 20 most occurred, strongly linked, and high-impact keywords. Future researchers use highly cited keywords as they receive global attention very fast as compared to least cited keywords. Keywords such as “COVID-19,” “Bioinformatics,” and “SARS-Cov-2” should be used along with “Proteinaceous,” “Peru,” “Screening,” “Oxidative stress,” etc., to explore the new facts and relationships. Countries like India, Nepal, and Bangladesh (Asian countries) have linkages with the “COVID-19” keyword. Bibliometric analysis can be an essential tool to detect research trends for the present and future (Pesta et al., 2018).

Figure 4. WordCloud of most frequent keywords (Software: Biblioshiny (R-studio)).

Author keywords co-occurrence analysis

Co-word analysis is the only technique that considers the publication’s content when measuring similarity while co-occurrence analysis, evaluates the papers more indirectly through citations (Kanta et al., 2021). In the present study, co-occurrence analysis is used to evaluate the articles published overtime based on author keywords. The fine-grained tropical structure is better understood by combining the keywords and cited references in the research field which also helps develop the relationships between various topics and their sub-topics (Van den Besselaar & Heimeriks, 2006). VOS-viewer software helps identify the themes using the keywords co-occurrence analysis in the specific study area (van Eck & Waltman, 2020), resulting in bibliographic clusters and emerging and least explored themes (Donthu et al., 2020).

1414 author keywords were found in the metadata extracted from the Scopus database. Only 78 keywords met the desired threshold when the minimum number of co-occurrences was restricted to 15 keywords. Figure 5 shows the visualisation of highly appeared keywords in various clusters marked in red (18 keywords), green (16 keywords), blue (13 keywords), yellow (12 keywords), purple (11 keywords), cyan (6 keywords), and orange (2 keywords) colours in the F1000Research journal since its inception. The red cluster is the most dominating and impactful cluster, whereas the orange is the least impactful cluster. Table 5 exhibits the 78 keywords that appeared in various clusters (depicted seven themes). These clusters emerged as different themes such as “bioinformatics” (69 occurrences), “treatment” (40 occurrences), “children” (33 occurrences), “COVID-19” (183 occurrences), “cancer” (66 occurrences), “inflammation” (35 occurrences) and “systematic review” (29 occurrences).

Figure 5. Author keywords co-occurrence and TLS visualisation (Data source: Scopus, Software: VOS-viewer).

In Figure 5, keywords are depicted by a node whose size specifies occurrences (van Eck & Waltman, 2020). Bigger the node size, the greater the occurrences of the keyword. Total link strength (TLS) has been represented by the line thickness between the nodes, representing the keyword’s co-occurrence frequency within the links exhibited in Table 5.

Cluster 1 (Red) = Bioinformatics. This cluster comprises 18 keywords and is the most significant among the 7 clusters. “Bioinformatics” was found as the maximum occurred keyword having 69 appearances and the maximum TLS of 65 exhibited in Table 5. It is well connected with “genomic,” “machine learning,” “RNA-Seq,” “R,” “Reproducibility,” “open science”, and so on. It shows that the F1000Research journal mainly covers medical science articles where the “bioinformatics” keyword plays a significant role. “Bioinformatics” is over 50 years old and recently emerged to support next-gen data analysis. It has faced multiple challenges recently while managing big data and the reproducibility of results, which can help integrate the same into academics (Gauthier et al., 2019). To extract information from big data, various machine-learning algorithms are widely used and applied in bioinformatics (Min et al., 2017). Machine learning has revolutionised computational biology (part of bioinformatics) in transforming massive data through technology into knowledge which can help understand genomic, proteomics and system biology (Larrañaga et al., 2006). This cluster discusses the science involved in handling bioinformatics and their application in improving education through software and increasing visibility through open science.

Cluster 2 (Green) = Treatment. This cluster comprises 16 keywords represented in Table 5 has shown “treatment” as a maximum occurred keyword (40 appearances) and TLS (40). It is well connected with various keywords such as “pregnancy,” “diagnosis,” “surgery,” “microbiome,” “tuberculosis,” “vaccine”, and so on. This cluster discusses the treatment of various diseases, infections, vaccines, and clinical trials. Pregnant women will continue to be a priority group for treatment optimisation in the era of compulsory treatment as treatment duration and antiretroviral therapy (ART) alternatives increase, promoting both their health and the health of their kids exposed to ART (Bailey et al., 2018). Diagnosis and treatment go hand in hand; therefore, these two words are in the same cluster.

Cluster 3 (blue) = Children. This cluster comprises 13 keywords represented in Table 5. The maximum occurred keyword is “children”, having 33 occurrences and a TLS of 22. TLS denotes that “children” is linked up with 22 different keywords. Significant linkages are “obesity,” “depression,” “COPD,” “stroke,” “elderly,” “mental health”, and so on. This cluster discussed the disease and its impact on mental health and well-being. Obesity has become a global health problem, becoming common due to increased screen time (World Health Organization, 2005). The advertisement of junk and fast food impacts children’s minds towards that food and ultimately results in obesity (Strasburger et al., 2011). Therefore, this cluster needs special attention.

Cluster 4 = COVID-19. This cluster has 12 keywords, and these are highly correlated to COVID-19. COVID-19 has the maximum number of occurrences, i.e., 183 and TLS of 165 in the F1000Research journal. None of the keywords have even 100 occurrences, whereas COVID-19 has 183, which shows the impact of COVID-19-related publications in the journal. “COVID-19” has been linked with “SARS-CoV-2,” “epidemiology,” “HIV” “risk factors,” “coronavirus,” “Prevalence,” “public health” “pandemic,” “Bangladesh,” “Nepal,” and “India” exhibited in Table 5. SARS-CoV-2 spread globally and was declared a pandemic by World Health Organisation. It damaged health and wealth and increased poverty globally. A review proposed by Ciotti et al. (2020) provided information regarding epidemiology, its origin, infection to humans and safety issues. Various researchers (Pokhrel & Chhetri, 2021) have also contributed to digitally improving the teaching-learning process to fight against a COVID-19-like pandemic in future. This cluster mainly focuses on the COVID-19 pandemic and its impact on developing countries like India, Bangladesh, and Nepal.

Cluster 5 = Cancer. The fifth cluster consists of 11 keywords, clustered around the “Cancer” keyword, 66 occurrences and 36 TLS exhibited in Table 5. “Case report,” “genetics,” “breast cancer,” “development,” and “immunotherapy” are the significant keywords strongly linked with the “cancer” keyword. This cluster mainly focuses on a widely spreading disease, “Cancer”, which must be prevented early rather than curing the advanced stage (Nixon et al., 1973). Due to the increase of cancer worldwide, cancer immunotherapy's impact has gained popularity in cancer clinical care (Blank et al., 2016). To combat this deadly disease, Silverstein et al. (2006) insisted on understanding the relationship between cancer and the immune system to develop treatment options for patients. Similar genomic profiles of tumors can help apply therapies to cancer types (Hegde & Chen, 2020).

Cluster 6 = Inflammation. This cluster comprises six keywords. The most impactful keyword in this cluster is “inflammation”, have 35 occurrences and a TLS of 25 exhibited in Table 5. The remaining five keywords, “pain,” “metabolism,” “mitochondria,” “diabetes,” and “oxidative stress,” are highly linked with the most impactful keyword, “inflammation.” This cluster tried to explain the symptoms and the diseases. Many chronic inflammatory diseases are caused due to imbalance of mitochondria, metabolism and inflammation (Tschopp, 2011). Inflammation can be a possible mechanism and prevent diabetes in type 1 and type 2 (Tsalamandris et al., 2019). Inflammation and oxidative stress are affected by hypertension and increase blood pressure regardless of medicine use (Pouvreau et al., 2018).

Cluster 7 = Systematic review. This is the smallest cluster that has only two keywords. The “Systematic review” keyword has 29 occurrences and 25 TLS, whereas “meta-analysis” has 24 occurrences and 20 TLS exhibited in Table 5. These two keywords are used as powerful tools for overcoming the handling of large-scale data. They can help present results from different studies conducted on a similar topic. A thorough understanding of meta-analysis is required to understand and accept the conclusions of various studies (Ahn & Kang, 2018).

Overlay visualisation of author keywords

Overlay visualisation of author keywords co-occurrences highlights the old and latest keywords through a bibliometric web map (van Eck & Waltman, 2010). Figure 6 exhibits the five clusters marked in five colours ranging from purple to yellow. Purple colours depicted the old keywords, such as “Genomics” “and Cytoscape,” whereas the latest keywords appeared yellow colour such as “COVID-19,” “Anxiety,” “Coronavirus,” “Obesity,” “Risk Factors,” and “Depression” are latest keywords. The latest and most popular keyword in 2020 was “COVID-19”; in the middle of 2019, the most popular keyword was “Machine learning.” It shows that COVID-19 fever is not yet over; researchers are still exploring this area as much as possible to fight against the deadly pandemic and to bring life back to normal.

Figure 6. Overlay visualisation of author keywords (Data source: Scopus, Software: VOS-viewer).

Documents-citation analysis

The citation analysis explains the highly cited documents and authors (Waltman et al., 2020) in the specific journal, performed through VOS-viewer software. Analysis was performed by restricting a document's minimum number of citations to 100. Out of 5518 documents, only 71 met the desired threshold. Table 6 exhibited the top 20 authors and articles with maximum citations in F1000 research journals. 71 documents with more than 100 citations which the previous researchers have used and now can be used in theoretical concepts for future studies. “Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences”, authored by Soneson et al. (2015), has a maximum citation of 1570, which is almost thrice more than the second highest cited document (574 citations), i.e., “FastQ Screen: A tool for multi-genome mapping and quality control” authored by Wingett & Andrews (2018). “Leishmaniasis: A review”, authored by Torres-Guerrero et al. (2017), has 516 citations, “Bioconductor workflow for microbiome data analysis: from raw reads to community analyses” authored by Callahan et al. (2016) has 439 citations and “Current understanding of Alzheimer’s disease diagnosis and treatment” authored by Weller & Budson (2018) has 434 citations. These documents can prepare a base for future researchers to develop a robust theory.

Most impactful authors

During the review of the F1000Research journal, 5290 authors were identified. While checking the citations analysis through VOS-viewer and limiting the maximum number of authors per document to 25 and the minimum number of documents of an author to two, only 225 authors met the desired threshold. Table 7 exhibits the top 20 authors, affiliations, country, total publications, total citations, and average cited documents since the journal’s evolution until 2022. Quantifying in publications is one of the criteria for increasing visibility on a global platform. However, it is challenging to gain popularity without citations of specific documents. Mainly European, Asian, and American continent authors’ visibility was found in the top-20 authors in terms of quantification of publication, whereas African and Australian authors did not appear in this list. The leading author’s impact is exhibited in Table 8 based on the h-index, g-index, and m-index. Various algorithms have been developed to calculate the most impactful author based on the publications, citations, author’s visibility, and continuity of publications. The g-indexed were designed to measure the global citation performance of articles or authors. It is also known as an improved version of the h-index (Egghe, 2006). The m-index is a metric used to measure the h-index divided by the start of the publication year until the latest year.

Table 8 exhibits the top 20 most impactful authors based on the h-index given by Hirsch (2005). F1000Research journal is one of the premium journals where 5290 authors have published their articles. However, only a few authors have more than a five h-index. “Gary D. Bader” and “Sean Ekins” have the maximum h-index of nine amongst the authors, 12 g-index, 12 publications and 577 and 269 citations, although the first publication came in 2014. “Jürgen Bajorath,” ranked number three in the list, has eight h-index, 19 publications, and has been active in publishing since 2012. Identifying the most prolific authors and their research articles can help future researchers extend their research recommendations, understand their research area, and identify the research gaps, which can help conceptualise the future research problem.

Bibliographic coupling of authors, countries, and organisations

Bibliographic coupling occurs when two papers cite a third common paper. The two papers address a common subject matter (Martyn, 1964). In the present study, the bibliographic links between authors, countries and organisations through overlay visualisation are shown in Figure 7, Figure 8, and Figure 9. While examining the bibliographic coupling of authors, the minimum number of documents of an author was restricted to 100. 5290 authors appeared, but only 225 met the desired threshold. However, the largest set of only 71 items connected items were found. Figure 7 depicts the bibliographic coupling between the authors at the various stages of a research journey. Author’s node marked in yellow highlights the youngest/recent (2021) coupling, whereas the authors in purple depicted the oldest (2016) coupling.

Figure 7. Bibliographic coupling between the authors (Overlay visualisation) (Data source: Scopus, Software: VOS-viewer).

Figure 8. Bibliographic coupling between the countries (Overlay visualisation) (Data source: Scopus, Software: VOS-viewer).

Figure 9. Bibliographic coupling between the organisations (Overlay visualisation) (Data source: Scopus, Software: VOS-viewer).

Figure 8 shows the bibliographic coupling (overlay visualisation) of countries. During the VOS-viewer bibliometric analysis, countries that published at least five documents were considered for final analysis. Only 89 countries met the desired threshold values out of 164 countries. However, the largest set of connected items was only 88. The United States was found as the most significant contributor in terms of publications as well as citations. They have been involved in publishing papers marked in purple for a long time. In contrast, Asian countries like Indonesia, United Arab Emirates and Qatar are new in bibliographic coupling (marked in yellow).

Figure 9 shows a bibliographic coupling (overlay visualisation) of the organisation. While performing the analysis using VOS-viewer, an organisation published at least four documents that were considered for the final analysis. Out of 14646 organisations, only 83 met the desired threshold criteria, but only 57 organisations had the largest set of connected items. Organisations marked in yellow are the youngest, and those marked in purple are the oldest in the F1000Research journal.

Emerging themes and trending topics

Keywords that appeared during the analysis in VOS-viewer software or Biblioshiny (R-studio) were visualised through theme generation. In the present study, Biblioshiny (R-studio) software was used to understand the various themes (niche theme, motor theme, basic theme and emerging or declining theme), which have been divided into four quadrants (Q1, Q2, Q3, and Q4) (Cobo et al., 2011), identified based on centrality (X-axis) and density (Y-axis). The degree to which a topic is connected to other topics and, in turn, significant in a particular domain is measured by centrality, which assesses the level of inter-cluster relationships. The density, conversely, gauges the degree of intra-cluster cohesion, or more specifically, how closely related the keywords in a given cluster are to one another and how strongly a theme is established (Forliano et al., 2021). In Figure 10, the upper left (high density and low centrality) includes niche research themes containing the keywords “open science,” “treatment,” “inflammation,” and “diagnosis.” Niche themes suggest it is internally well developed but unable to influence others due to low centrality. Motor themes appeared in the upper right quadrant (high in density and centrality) and included the keywords “bioinformatics,” “genomics,” and “RNA-seq.” It suggests that themes are well-developed and highly influence the researcher. Basic themes appeared in the lower right quadrant (high in centrality and low in density), showing the themes are extending or lying across for discipline and can influence the other researcher/topics but are underdeveloped. Emerging or declining themes appeared in the bottom left quadrant (low centrality and density), which is neither well developed nor influenced by the researcher. Keywords that appeared here are “COVID-19,” “SARS-CoV-2,” and “Children” are a matter of great concern in the present scenario. Based on the thematic map analysis, it can be concluded that the keywords that appeared in niche themes are well-developed and highly influenced by the researchers. In contrast, future researchers need more focus on emerging or declining themes to develop a concrete plan to fight against these widespread (COVID-19, SARS-CoV-2, and children) keywords.

Figure 10. Emerging theme identification (Data source: Scopus, Software: Biblioshiny – R Studio).

The present study visualises trending topics through Biblioshiny (R-studio), shown in Figure 11. Bubbles of the smallest size have shown a minimum of 50, middle 100, and biggest 150 appearances. Recently appeared topics (between 2021-22) are awareness, attitude and COVID-19, although their bubble size is small (minimum <50 and >100 appearances). COVID-19 again appeared between 2020-21 with the biggest bubble (<150 appearances) along with “SARS-Cov-2” (<100 appearances) and “systematic review” keywords. These topics emerged based on keywords in the author’s articles published in the F1000Research journal journey since its inception.

Figure 11. Trending topics (Data source: Scopus, Software: Biblioshiny – R Studio).