Leveraging H3Africa Scholarly Publications for Technology-Enhanced Personalized Bioinformatics Education

: The Coronavirus Disease 2019 (COVID-19) pandemic has catalyzed the expectations for technology-enhanced interactions with personalized educational materials. Adjusting the content of educational materials to the geographical location of a learner is a customization feature of personalized education and is used to develop the interest of a learner in the content. The educational content of interest in this report is bioinformatics, in which the knowledge spans biological science and applied mathematics disciplines. The Human Heredity and Health in Africa (H3Africa) Initiative is a resource suitable for use when obtaining data and peer-reviewed scholarly articles, which are geographically relevant and focus on authentic problem solving in the human health domain. We developed a computerized platform of interactive visual representations of curated bioinformatics datasets from H3Africa projects, which also supports customization, individualization and adaptation features of personalized education. We obtained evidence for the positive effect size and acceptable usability of a visual analytics resource designed for the retrieval-based learning of facts on functional impacts of genomic sequence variants. We conclude that technology-enhanced personalized bioinformatics educational interventions have implications in (1) the meaningful learning of bioinformatics; (2) stimulating additional student interest in bioinformatics; and (2) improving the accessibility of bioinformatics education to non-bioinformaticians.


Introduction
The Coronavirus Disease 2019 (COVID-19) pandemic has challenged traditional methods of design, delivery and assessment aspects of education globally [1][2][3][4]. Additionally, the COVID-19 pandemic has catalyzed the expectations for technology-enhanced interactions with personalized educational materials [5][6][7]. Adjusting the content of educational materials to the geographical location of a learner is a customization feature of personalized education [8] and is used to develop the interest of a learner in the content [9,10]. The educational content knowledge of interest in this report is bioinformatics, the application of tools of computation and analysis to the capture and interpretation of biological data [11]. Bioinformatics education has been broadly defined as "the teaching and learning of the use of computer and information technology, along with mathematical and statistical analysis for gathering, storing, analyzing, interpreting, and integrating data to solve biological problems". Thus, bioinformatics knowledge spans biological and applied mathematics disciplines [12]. Globally, bioinformatics education exists at multiple academic levels and is an indispensable part of life science education [13][14][15][16]. Genomic sequence data from different parts of the world are also available as curriculum resources for bioinformatics education [13]. The global aspects of bioinformatics present opportunities for customizing bioinformatics educational resources through the geographic context of the bioinformatics data. For example, databases on human genomic variants include data on geographic regions to facilitate interpretation [17,18].
Advances in multiple research fields, including instructional design and visual analytics, have high potential to influence educational practices in the field of bioinformatics. The instructional design features of personalized education include customization (the system supports an individual's choice), individualization (the individual makes the choice) and adaptation (the system makes the choice) [8]. Personalized education promotes creativity and innovation, which are needed in a modern society [19]. Visual analytics software, which belongs to the software component of applied mathematics [12], enables the design of visualizations for the representation, interaction and analysis of data [20]. The purpose of this report is to describe a computerized platform of interactive visual representations of curated bioinformatics datasets to support features of personalized education, namely customization, individualization and adaptation. Our perspective is that digital software technology platforms can especially enable students and non-specialists to visually analyze and interpret bioinformatics data.
The Human Heredity and Health in Africa (H3Africa) Initiative was established in 2012 with a vision that the data generated from a pan-continental network of laboratories will inform strategies to ultimately lead to health benefits in Africa [21,22]. H3Africa is a resource suitable for obtaining data and peer-reviewed scholarly articles that are geographically relevant and focuses on authentic problem solving in the human health domain [21,22]. In 2022, after 10 years of the establishment and development of H3Africa, several bioinformatics advances have helped to accomplish the H3Africa vision [23][24][25]. As of June 2022, advances in the generation and investigation of data from genome and genome-enabled resources have been documented in at least 615 scholarly publications [26]. Peer-reviewed scholarly publications contain textual content, images, data tables and supplementary materials that are suitable for use in technology-enhanced educational resources to acquire knowledge for performing bioinformatics tasks. Additionally, since genomic data generation is a major component of H3Africa projects [27], we expect that H3Africa publications will include bioinformatics knowledge categories that are described in bioinformatics journals.
Bioinformatics problems can be complex and ill-structured, requiring the application of multiple bioinformatics knowledge categories [28][29][30]. The network of bioinformatics knowledge includes bioimage informatics; data and text mining; databases and ontologies; gene expression; genetics and population analysis; genome analysis; phylogenetics; sequence analysis; structural bioinformatics; and systems biology [31,32]. The complexity of bioinformatics knowledge brings to relevance the need for meaningful learning, in which the learner builds mental models (representations) by combining new knowledge and any relevant prior knowledge with opportunities to test mental models through problems, refined by experience and the appropriate feedback [33][34][35][36]. The five dimensions of meaningful learning are active, authentic, constructive, and cooperative and personalized [36]. Meaningful learning can help accomplish the educational goals of the long-term retention of knowledge and the transfer of knowledge to new problems [35,37]. In technology-enhanced learning environments, interaction facilitates meaningful learning [38], and the design of interfaces for effortful interaction can support the retrieval of relevant knowledge from multiple options. In the learning strategy of retrieval practice, knowledge is retrieved from memory to answer questions such as short answer and multiple-choice formats [39]. We envision that in a visual analytics software, the retrieval of knowledge can be achieved by interaction and responses provided to a quizzing software. We therefore used this approach to determine the effectiveness of one of the visual analytics resources designed to support the learning of facts about genomic sequence variations. The design of interfaces for learning facts needed in bioinformatics problem solving can therefore benefit from incorporating mental modeling stages, retrieval practice and the dimensions of meaningful learning.
A collection of peer-reviewed H3Africa scholarly publications from 2012 provides both historical and emerging knowledge useful in bioinformatics educational activities including technology-enhanced design, delivery and assessment. In summary, the growing number of H3Africa publications can be a suitable resource for designing personalized bioinformatics education that includes customization (e.g., by geographic location), individualization (e.g., choice of population groups) and/or adaptation (e.g., collecting data on visual analytics actions). In this article, we present how we reused the content of an open access research article entitled "High-depth African genomes inform human migration and health" [40]. The major results include recoded and reshaped datasets as well as designs of visual analytics resources that can support the personalized education of genome-level population variation in Africa. A pilot evaluation study for effectiveness and perceived usability resulted in a positive effect and acceptable usability.

Materials and Methods
From September 2017 to June 2022, the development of visual analytics resources to support bioinformatics learning was an activity in the Train-the-Trainer Project of the H3Africa Bioinformatics Network (H3ABioNet). We designed and implemented the visual analytics resources with Tableau, and they were categorized as worksheet (a template for designing representations, interactions and analyses), dashboard (combining worksheets and other objects) and story (combining worksheets and dashboards). Several visual analytics resources are available at https://public.tableau.com/app/profile/uibgnode (accessed on 12 August 2022). We Similar to our previous publication on sickle cell disease [41], we reused the textual content of the article and associated supplementary resources as the data sources for the design of visual analytics resources. In this report, we recorded or reshaped datasets from an H3Africa scholarly article to render it machine-readable and facilitate the design of the personalized education of genome-level variation in Africa. The method sub-sections below describe approaches that we used to leverage the H3Africa open access publications to design a computerized platform to support personalized bioinformatics education.

Recoding and Reshaping Datasets from an H3Africa Scholarly Article
The text of the scholarly article and the supplementary spreadsheet file of the PubMed Central (PMC) article PMC7759466 were the data sources for the construction of value-added, visual-analytics-ready datasets. For a dataset in a worksheet of a supplementary spreadsheet file, we recoded and or reshaped the dataset to be machine-readable, in this case for use in visual analytics software. The recoding included adding annotations (or labels) and unique identifier column for each record (row) in the dataset. We may have also added a column that provides notes about entries in the record. For example, the Supplementary Table 6 of PMC7759466 is a color-coded dataset for the detection of genes showing outlier Composite Likelihood Ratio (CLR) scores (p < 0.001) in six populations. We recoded the dataset that encoded the detection of a gene in a population as "1" when detected and "0" when not detected ( Figure 1). In the visual analytics software (Tableau Desktop or Tableau Public) [42], we implemented visual analytics actions such as search and filter as well as a constructed a six-digit binary pattern from the six populations for each gene entry. Table 6 of PMC7759466 is a color-coded dataset for the detection of genes showing outlier Composite Likelihood Ratio (CLR) scores (p < 0.001) in six populations. We recoded the dataset that encoded the detection of a gene in a population as "1" when detected and "0" when not detected ( Figure 1). In the visual analytics software (Tableau Desktop or Tableau Public) [42], we implemented visual analytics actions such as search and filter as well as a constructed a six-digit binary pattern from the six populations for each gene entry.  Table 6 in PMC7759466. In the middle image, a section of the value-added and visual-analytics-ready dataset is shown. The first column provides the value of  Table 6 in PMC7759466. In the middle image, a section of the value-added and visual-analytics-ready dataset is shown. The first column provides the value of unique identifier for each record in the constructed dataset. The binary digit entries of "0" and "1" provide value to encode the detection of the gene (second data column) in six populations: multiethnic from Botswana (BOT), Bantu Speakers from Zambia (BSZ), Bantu and Bantoid Speakers from Cameroon (CAM), Berom from Nigeria (BRN), Fon in Benin (FNB) and Gur speakers from West Africa (WGR). The last data column encodes the number of populations in which the gene was detected. In the bottom image, the binary digits are combined in visual analytics software to accomplish visual analytics actions such as filtering and connection to an external website. The visual analytics workbooks are available for download at https://public.tableau.com/app/profile/ uibgnode/viz/pmc7759466_analytics/overview (accessed on 12 August 2022).

Supplementary
We also recoded a dataset of textual statements by copying sentences from the PMC article to a worksheet of spreadsheet software (e.g., Microsoft Excel and Google Sheets). Since sentences may express a complete thought [43], we extracted sentences to support reading strategies (such as skimming and scanning) in visual analytics software. Each of the statement column entries had associated data columns with the section (e.g., abstract and methods) and subsection source of the statement. An additional approach was to obtain the visual-analytic-ready json format of PubMed Central article using the uniform resource locator script: https://www.ncbi.nlm.nih.gov/research/bionlp/RESTful/pmcoa. cgi/BioC_json/PMC7759466/unicode (accessed on 12 August 2022).

Design of Visual Analytics Resources for Personalized Learning of Bioinformatics Knowledge
We accomplished the design of the visual analytics resources in the following two main stages. In the first stage, we imported relevant dataset(s) into Tableau [42]. In the second stage, we arranged the data fields from the datasets in worksheets or dashboards to accomplish the desired personalization. The arrangement of the data fields followed guidelines for using interactive visual representations to support learning and other complex activities [20,44,45]. The designs included features such as filters and connections to websites of bioinformatics resources and scholarly publications. These design features are intended to support the effortful and intentional interaction with datasets to achieve personalization, namely, customization, individualization and adaptation [8].

Evaluation of Visual Analytics Resources
We used a pre-test and post-test quasi-experimental design to determine the effect size of using the interaction with a visual analytics resource to obtain answers to questions. This method enables effortful interaction that leads to the retrieval of the answers in the visual analytics resource. This form of retrieval practice of facts knowledge promotes meaningful learning [35,37], where the accumulation of facts leads to learning with understanding [33]. The visual analytics resource that we evaluated was designed to support learning of the impact (low, moderate, modifier or high) of genomic sequence variants on transcript structure or protein function [46,47]. We informed the participants about the study procedures, which consisted of responding to four identical multiple-choice questions before and after interacting with a visual analytics resource for 10 to 15 min. We present the four questions used in the study in Table A1 of Appendix A. The consenting respondents were participants in a bioinformatics or bioinformatics-related educational activity in a higher education setting. We obtained the mean difference effect size of the unpaired groups (pre and post test scores) using estimation statistics and visualized the comparison using a Gardner-Altman estimation plot [48].
We also used the Systems Usability Scale (SUS) questionnaire to measure the perceived usability of the selected visual analytics resource [49,50]. Respondents completed the SUS questionnaire after interacting with the visual analytics resource. The SUS questionnaire consists of 10 questions, and the scoring system ranges from 0 to 100. The SUS can provide the reliable measurement of the perceived usability of a system using data from 8 to 12 respondents [50]. We used the web-based System Usability Scale Analysis Toolkit to analyze and generate visualizations of the SUS data [51]. The SUS questions adapted to use for the visual analytics resource are shown in Table A2 in Appendix A. In the sections below, we describe use cases of the visual analytics designs that support customization, individualization and adaptation in technology-enhanced personalized education. In each use case, we provide the following descriptions: (1) principle(s) of the personalization design(s); and (2) design objective and results. • Individualization Design Objective and Results. The top image of Figure 2 is an example of individualization, when the learner enters the gene symbol RNF135 in the NCBI search feature and returns 21 databases. The bottom image in Figure 2 is the result when the learner interacts with the visual analytics resource in the top image of Figure 2 and makes the choice to display key genes relevant to a population (BRN) in Nigeria. The result of the interaction includes two GO terms. The learner is interested in the pancreatic lipase gene (PNLIP) based on the NCBI search.

A Use Case of Adaptation in Personalized Learning by Collecting Performance Metrics during Actions in Visual Analytics Software
• Principle of Adaptation Design in Personalized Educational Technology. In adaptation design in personalized education technology, the system makes the choice based on data patterns such as repeated behaviors and performance measures [8]. The adaptation design can have a diagnostic function to optimize the learning transaction strategies of the learner. The adaptation design requires the completion of both customization and individualization activities. • Adaptation Objective, Procedure and Results. The bioinformatics content for this use case is genomic variants, which are changes (insertions, deletions or replacements) in one allele of a gene of an individual compared with a reference genome [53,54]. There are at least 39 calculated consequences of genomic variations on transcripts or proteins that are located on the transcript structure [55]. The customization design in Figure 3 is a dashboard that consists of (1) a sequence ontology description and other descriptive features for sequence variants; (2) an image from the Ensembl website illustrating the location of sequence variants in the gene structure; (3) filters for specifying the sequence ontology descriptors. We applied the performance-recording function of Tableau Desktop (accessed via Help menu) to acquire data patterns of events performed by software in response to an individual's requests ( Figure 4). Patterns reported include the blending of data, computing layout, compile query, executing query, rendering, computing layout and sorting data. In Figure 4, the differences in the amount of interaction are displayed using performance metrics available in Tableau Desktop Professional [42]. The top image shows activities by the visual analytics resource that reflects that the learner did not make changes to the filters during the period of the interaction. In the bottom image, the performance metrics reflect that the learner changed the value of the impact and included text in the sequence ontology description.  The design provides opportunities to read text and view images on the descriptions of sequence variants. In addition, the individual is able to compare the definitions of the sequence variants. Personalized learning on sequence variants could start with this dashboard to learn concepts (content types that can be defined) and then other visual analytics sheets on sequence variants for data investigations. The image was obtained from Ensembl Variation-Calculated variant consequences (https://www.ensembl.org/info/genome/variation/prediction/predicted_data.html) (accessed on 12 August 2022) [55]. The visual analytics workbook is available for download at https:// public.tableau.com/app/profile/uibgnode/viz/pmc7759466_suppT19_variants/overview (accessed on 12 August 2022). Figure 3. A visual analytics dashboard designed for personalized learning of the concept of functional consequences of sequence variants. The design provides opportunities to read text and view images on the descriptions of sequence variants. In addition, the individual is able to compare the definitions of the sequence variants. Personalized learning on sequence variants could start with this dashboard to learn concepts (content types that can be defined) and then other visual analytics sheets on sequence variants for data investigations. The image was obtained from Ensembl Variation-Calculated variant consequences (https://www.ensembl.org/info/genome/variation/prediction/predicted_data.html) (accessed on 12 August 2022) [55]. The visual analytics workbook is available for download at https://public.tableau.com/app/profile/uibgnode/viz/pmc7759466_suppT19_variants/overview (accessed on 12 August 2022).

•
Principles of Reading Strategies. The reading strategies of skimming (seeking main ideas) and scanning (keyword search) can support the efficient and effective comprehension of online scholarly publications [56]. The growing number and expanding bioinformatics content of H3Africa scholarly publications [27] presents opportunities in personalized bioinformatics education to support readers of H3Africa publications to use reading strategies such as the skimming and scanning of knowledge (content) types [57]. When skimming, the reader seeks the main idea using the outline or table of contents in the text.

•
Reading Strategies Objective and Results. The objective of this use case on reading strategies is to demonstrate skimming and scanning reading strategies with a textual dataset. We used the recoded dataset of 381 statements from the H3Africa publication PMC7759466 [40], which included section headings, statement text and statement identifiers. We designed visual analytics worksheets to support skimming and scanning ( Figure 5). When skimming, the reader can display the list of section headings to obtain the main ideas described in the article. The support for scanning allows the reader to search by keywords and retrieve statements that contain keywords as well as details on the context of use in one or more instances. The statement(s) retrieved can be the basis for an intensive reading strategy where the purpose is to obtain the exact understanding of the text [58].
Educ. Sci. 2022, 12, x FOR PEER REVIEW 11 of Figure 5. Examples of support for reading strategies of skimming and scanning for efficient a effective reading of H3Africa publication on genome-level variation in African populations. The t image shows visual analytics support for skimming the outline of the article. A subset of the bio formatics methods is shown. The bottom image shows visual analytics support for scanning. T four statements that contain "sickle cell" are shown. The retrieved statements on sickle cell can read to understand the use of diverse frameworks including the five content types, namely, fa concept, process, procedure, and principle [59]. For example, statement RA_368 in the figure d scribes the concept of sickle cell anemia by the associated genome-level variation in missense mu tion in the HBB (hemoglobin subunit beta) gene. The four statements that contain "sickle cell" are shown. The retrieved statements on sickle cell can be read to understand the use of diverse frameworks including the five content types, namely, fact, concept, process, procedure, and principle [59]. For example, statement RA_368 in the figure describes the concept of sickle cell anemia by the associated genome-level variation in missense mutation in the HBB (hemoglobin subunit beta) gene.

Evaluation of Visual Analytics Resources
Our initial evaluation of visual analytics resources focused on a visual analytics resource designed to support the personalized learning of genomic sequence variants (Figure 3). The four questions for pre-test and post-test required the selection of the correct example of sequence variant impact in categories of low, moderate, modifier and high. Thirty-eight respondents completed the pre and post surveys. We determined the effect size measured by the mean difference of the two categories of scores (pre-test and post-test) for the 38 respondents (Figure 6a). We calculated the unpaired mean difference between pretest scores and post-test scores as 1.29 [95.0%CI 0.789, 1.74] using the web-based estimation statistics software [48]. The p value of the two-sided permutation t-test was 0.0.

Evaluation of Visual Analytics Resources
Our initial evaluation of visual analytics resources focused on a visual analytics resource designed to support the personalized learning of genomic sequence variants (Figure 3). The four questions for pre-test and post-test required the selection of the correct example of sequence variant impact in categories of low, moderate, modifier and high. Thirty-eight respondents completed the pre and post surveys. We determined the effect size measured by the mean difference of the two categories of scores (pre-test and posttest) for the 38 respondents (Figure 6a). We calculated the unpaired mean difference between pre-test scores and post-test scores as 1.29 [95.0%CI 0.789, 1.74] using the web-based estimation statistics software [48]. The p value of the two-sided permutation t-test was 0.0.
The analysis of the responses of nine respondents to the 10 questions from the System Usability Scale questionnaire generated a SUS Study Score of 74.17, a median of 80 and standard deviation of 12.91. The perceived usability was acceptable according to the webbased SUS Analysis Toolkit (Figure 6b) [51]. Score is shown in the above Gardner-Altman estimation plot. Both groups are plotted on the left axes; the mean difference is plotted on a floating axis on the right as a bootstrap sampling distribution. The mean difference is depicted as a dot; the 95% confidence interval is indicated by the ends of the vertical error bar. In the System Usability Score image (b), the dots are the scores provided the respondents, and the distribution of the scores is illustrated as a box plot. The mean SUS score of 74.1 is within the acceptable range of usability. Figure 6. Evidence for positive effect size and acceptable usability of a visual analytics resource designed to support personalized education of the concept of functional consequences of sequence variants. In the effect size image, (a) the mean difference between Pre-Test Score and Post-Test Score is shown in the above Gardner-Altman estimation plot. Both groups are plotted on the left axes; the mean difference is plotted on a floating axis on the right as a bootstrap sampling distribution. The mean difference is depicted as a dot; the 95% confidence interval is indicated by the ends of the vertical error bar. In the System Usability Score image (b), the dots are the scores provided the respondents, and the distribution of the scores is illustrated as a box plot. The mean SUS score of 74.1 is within the acceptable range of usability.

Discussion
The analysis of the responses of nine respondents to the 10 questions from the System Usability Scale questionnaire generated a SUS Study Score of 74.17, a median of 80 and standard deviation of 12.91. The perceived usability was acceptable according to the web-based SUS Analysis Toolkit (Figure 6b) [51].

Discussion
Findings from systematic reviews have concluded that "personalized education is an attempt to improve the diagnosis, prediction, and treatment of learning outcomes alongside the prevention of learning losses" [60]. Learning outcomes in educational settings include long-term retention, the transfer of learning and desire/motivation for future learning or enrollment [37,[61][62][63]. These learning outcomes are also relevant in bioinformatics education [30]. We implemented a research and design framework for personalized bioinformatics education that integrates features of customization, individualization and adaptation within a visual analytics software platform [8]. We selected visual analytics software as the digital platform technology for implementing personalized education because of the enormous and fast pattern perception capability of the human visual system [64]. Additionally, visual analytics software enables interaction designs (e.g., arranging, filtering, scrolling, selecting and sorting) [45], which support learner transactions with bioinformatics facts, concepts, process, procedures and principles. Visual analytics software supports speed-of-thought interaction and a sense of immediacy that reduces the time needed for knowledge generation [65,66]. We applied the instructional principles of five content types to support meaningful or robust learning outcomes such as long-term retention and learning transfer [59,61,62]. We identified the collection of H3Africa scholarly bioinformatics articles as a source of geographically relevant bioinformatics knowledge. There is an opportunity to incorporate and integrate cultural knowledge, technology and locational context to bioinformatics education in Africa, as recently demonstrated in the meaningful learning of chemistry in Africa [67].
Approaches to learner interest development in bioinformatics content are a priority to address the needs of the bioinformatics workforce [30]. Interest is "a person's psychological state during engagement with some content, as well as the motivation to seek information and reengage with that content over time [68]." Personalized educational interventions have helped to increase the interest of learners [9]. We propose that incorporating the four stages of interest development (triggered situational interest, maintained situational interest, emerging individual interest and well-developed individual interest) [68] to bioinformatics education could help learners make meaningful connections to the bioinformatics content. H3Africa publications are authentic (real-world) Africa-centric educational resources, which can elicit situational interest, a heightened state of attention and increased engagement that could lead to the emergence of enduring individual interest [9]. Bioinformatics educators could use the curated datasets and visual analytics resources for students to seek bioinformatics information and engage with the bioinformatics content over time. Areas of future research could be how sustained engagement with authentic bioinformatics content help students to develop (1) confidence to investigate bioinformatics datasets; (2) bioinformatics disciplinary identity; and (3) bioinformatics careers intentions.
We have produced a unique set of twenty-five curated bioinformatics datasets from a major publication of the H3Africa consortium [40]. In reading processes, the human visual system detects and recognizes representations or patterns in image and grammatical units such as words, sentences and paragraphs. We designed the four visual analytics workbooks to include sheets with interactive tabular formats for the customization and individualization of personalized education. Learner performance is enhanced when interactions occur with appropriate representations [69]. Therefore, the data resources produced combined with the H3Africa article can be curriculum materials for diverse content, including human migration and genomics in populations in Africa. Engaging with the primary scientific literature through appropriate reading strategies and domain knowledge is an essential skill for successful careers in scientific fields, including bioinformatics [70,71]. The H3Africa publications include scholarly articles that are dense with bioinformatics content types and could be challenging for emerging researchers. A potential solution to alleviate the demands (such as high cognitive load and time) during the reading of bioinformatics is the visual-analytics-based presentation of journal article text combined with interpretation frameworks of the five content types [59]; the biological and applied mathematics landscapes [12] and the data-information-evidence-knowledge sequence [72]. We obtained evidence of the positive effect size and acceptable usability ( Figure 6) of a visual analytics resource designed to support the personalized learning of the concept of functional consequences of sequence variants. The system usability scale [50,51] can be a rapid evaluation strategy of the visual analytics resources and the datasets that we developed. We propose that this set of questions can be used to evaluate the usability of the visual analytics resources reported in this article. The availability of web-based toolkits (such as https://mixality.de/sus-analysis-toolkit/ (accessed on 12 August 2022) [51] can help provide educators with feedback on the usability of the visual analytics resources and inform improvements.
Regarding the adaptation feature of personalized education, in Figure 4, the potential use of the performance recording option in Tableau software is illustrated [73]. We envision a potential use in personalized education, where the learner or learning facilitator can determine, from the timeline of events, the level of complexity of interactions and potential complex cognitive activities [45]. These computational measurements could be further investigated as a proxy for measuring the complexity of activities employed when interacting, representing or analyzing bioinformatics content in visual analytics software. We encourage additional research on how to interpret these performance measures. Sharing the learning performance can be helpful to learners and help learning facilitators to provide feedback on visual analytics solutions to data investigation problems.
The visual analytics resources produced by our research efforts have been disseminated via the Tableau Public website (https://public.tableau.com/app/profile/uibgnode (accessed on 12 August 2022). This dissemination allows individuals to acquire competencies for working with software that is general purpose and utilized in the workplace [74]. A potential limitation to the use of the visual analytics sheets is software costs. There are several approaches to overcome these concerns in educational settings, including free license keys for students and instructors. The free online Tableau Public website has online authoring. The views in the Tableau workbooks can be interacted with offline using the free Tableau Reader (https://www.tableau.com/products/reader (accessed on 12 August 2022)). Individuals or research teams with licenses for the Tableau Desktop Professional can modify the visual analytics design, work offline and incorporate their datasets of interest to the visual analytics workbooks. This approach could also be applied using other visual analytics or business intelligence tools [75].

Conclusions
We recoded, reshaped and visually represented textual and numerical datasets from a high-impact publication (PubMed Central Identifier: PMC7759466) on genomics in Africa. The designs of the visual analytics resources described supports features of personalized education including customization, individualization and adaptation. We conclude that technology-enhanced personalized bioinformatics educational interventions has implications in (1) the meaningful learning of bioinformatics; (2) stimulating additional student interest in bioinformatics; and (3) improving the accessibility of bioinformatics education to non-bioinformaticians.