African Viral Dark Matter: From Viral Discovery to Anticipatory Surveillance

1. Introduction

Pandemic preparedness remains largely organized around known pathogens. Surveillance systems typically begin with a predefined pathogen list, diagnostic target, reference genome, and epidemiological hypothesis. This model is efficient when the causative agent is recognized, but it becomes intrinsically limited when a virus is divergent, previously undescribed, or absent from reference databases.

Metagenomic sequencing offers a different starting point. Rather than testing whether a specimen contains one of a predefined set of pathogens, metagenomics interrogates its nucleic-acid content without requiring a priori knowledge of the target. This capability has expanded its use in unexplained infections, outbreak investigation, and emerging-virus surveillance [9,17].

Yet metagenomics does not eliminate the unknown. It relocates it.

The unknown becomes a problem of detection, assembly, classification, and biological interpretation. Sequences may fail to assemble, lack recognizable homologues, be misclassified as cellular DNA, derive from endogenous viral elements, reflect environmental contamination, or represent viruses whose biological significance cannot be inferred from sequence data alone. Viral dark matter therefore arises from the interaction of biological diversity, sampling, sequencing technology, computational methods, reference databases, and interpretation [1,2,3,4].

The scale of this problem is substantial. Recent viromics studies continue to identify large numbers of viral sequences with no recognizable homologues or functional annotation, while large-scale catalogues of uncultivated viral genomes demonstrate that database absence does not imply biological rarity [3,4,5]. Viral dark matter is consequently not synonymous with pathogenicity or even with novelty in the strict taxonomic sense. It is, fundamentally, an epistemic limitation in what current surveillance systems can observe and interpret.

This limitation has particular relevance in Africa. The continent encompasses tropical forests, savannahs, wetlands, deserts, coastal ecosystems, agricultural landscapes, and rapidly expanding urban centres. These environments support dynamic interactions among humans, livestock, wildlife, arthropod vectors, and environmental reservoirs. Land-use change, agricultural expansion, urbanization, wildlife trade, livestock production, and changes in vector distributions can further modify these interfaces.

At the same time, viral reference space remains uneven. Recent African metagenomic studies have identified previously uncharacterized viral diversity in patients with acute febrile illness, mosquitoes, bats, wild small mammals, and human–animal interfaces [9,10,11,12,13,14,15,16]. These findings demonstrate that conventional surveillance captures only a fraction of the viral diversity present in sampled African populations.

Crucially, these empirical signals do not demonstrate that Africa contains a disproportionate burden of dangerous unknown viruses. Instead, as visually framed by the structural gaps in Figure 6, they reveal a structural knowledge asymmetry: the viral diversity observed across African ecosystems represents a fraction of the diversity that remains unresolved by current sampling and analytical systems.

We use the term African Viral Dark Matter to denote this structural knowledge asymmetry. The term does not imply an intrinsically elevated regional biosecurity risk. It describes viral diversity that remains undetected, incompletely characterized, or biologically uninterpretable within existing surveillance boundaries.

The central analytical challenge is therefore not simply to discover more viral sequences. It is to determine which signals represent biologically meaningful viral populations, where they occur, which hosts and ecological compartments they involve, how their distribution changes over time, and whether those changes warrant intensified surveillance.

Here, we propose a conceptual framework that moves from viral discovery to viral intelligence. We distinguish demonstrated observations from ecological plausibility and unresolved biological questions using an Evidence–Plausibility–Unknowns (EPU) framework. We then propose an African Viral Dark Matter Atlas (AVDMA) and a Viral Relevance and Priority Score (VRPS) to support longitudinal, context-rich, and transparent prioritization. Importantly, we advocate for anticipatory surveillance rather than deterministic prediction. The objective is not to identify the next pandemic virus in advance, but to systematically reduce uncertainty before an outbreak, identify changes in viral systems, and shorten the interval between ecological perturbation and epidemiological recognition.

2. From Pathogen Surveillance to Viral Discovery

Conventional surveillance is optimized for recognition; viral discovery is optimized for exploration. The distinction is fundamental. A conventional diagnostic workflow may test a patient with fever for malaria, dengue, chikungunya, influenza, SARS-CoV-2, or other locally relevant pathogens. Such testing remains indispensable because it is rapid, standardized, and clinically actionable. A negative panel, however, does not establish the absence of a viral cause; it indicates only that the tested diagnostic hypotheses were not supported.

Clinical metagenomics changes this logic by allowing pathogen-agnostic interrogation of clinical specimens, as illustrated in Figure 1. Its application to unexplained acute febrile illness in Uganda illustrates this potential: metagenomic sequencing was used precisely because conventional approaches could not resolve a substantial proportion of cases [9]. More broadly, clinical metagenomics is increasingly being evaluated for diagnosis, outbreak investigation, and surveillance of viral pathogens [17].

But pathogen-agnostic sequencing should not be confused with pathogen-agnostic interpretation. A detected sequence may represent a clinically relevant virus, a passenger virus, a transient environmental signal, a contaminant, a defective genome, an endogenous viral element, or a virus replicating without causing disease. Consequently, the progression from sequence detection to virus identification, biological attribution, epidemiological significance, and public-health relevance requires progressively stronger evidence.

A divergent viral contig is therefore a scientific observation, not by itself evidence of a new pathogen. This distinction is especially important for discovery programmes in Africa. Increasing the number of viral genomes deposited in international databases is valuable, but sequence accumulation without ecological and biological context has limited capacity to inform surveillance. Viral intelligence requires information on geographic origin, host species, age and health status, ecological setting, seasonality, vector abundance, environmental conditions, longitudinal persistence, co-infections, evidence of replication, serological exposure, and, where appropriate, evidence of disease.

The conceptual transition is therefore: sequence detection → authentication → taxonomic placement → host association → ecological persistence → biological activity → transmission evidence → disease association → public-health relevance. Each step reduces uncertainty, but none can be assumed from the preceding step.

3. Why African Viral Dark Matter Remains Undercharacterized

The African Viral Dark Matter concept rests on a simple observation: the distribution of known viral sequences is shaped by the spatial and institutional distribution of sampling and analytical capacity. Three interacting asymmetries are particularly important.

3.1. Sampling Asymmetry

Sampling determines what can be observed. Human surveillance has historically focused on symptomatic patients and recognized clinical syndromes. Animal surveillance often concentrates on economically important livestock or previously recognized reservoir species. Vector surveillance is frequently organized around established arboviruses. Environmental viromes remain substantially less characterized. This creates a predictable observation bias: viral diversity in rarely sampled hosts, locations, or ecological compartments remains invisible.

Recent metagenomic investigations illustrate this principle. Studies of acute febrile illness in Uganda [9], patients in Nigeria [10,11], mosquitoes in Kenya [12] and Burkina Faso [13], mosquitoes in South Africa [14], wild small mammals in West Africa [15], and human–animal interfaces in Ghana [16] have identified viral diversity not adequately represented by conventional surveillance.

Evidence: These studies demonstrate that previously uncharacterized viral sequences and lineages can be detected across diverse African human, wildlife, and vector-associated samples. They provide direct evidence that existing surveillance does not capture the full viral diversity of sampled ecosystems.

Plausibility: Differences in sampling intensity, host selection, geographical coverage, and ecological representation plausibly generate systematic blind spots. Viral populations associated with unsampled hosts or ecological compartments are less likely to enter reference databases or surveillance systems.

Unknowns: The magnitude of this blind spot remains unresolved. Current studies cannot establish the total viral diversity present across African ecosystems, nor determine what proportion of uncharacterized sequences correspond to active, persistent, or epidemiologically relevant viruses. The implication is methodological rather than geographical: the map of viral diversity partly follows the map of sampling effort.

3.2. Reference-Database Asymmetry

A sequence without a close reference may be classified as unknown even when it represents a biologically real virus. This produces a paradox: the more divergent a virus is from characterized viruses, the less likely conventional homology-based pipelines are to recognize it. Viral dark matter can therefore become greatest precisely where reference coverage is weakest.

VirSorter2 and related tools have improved the detection of diverse viral sequences, including groups poorly represented in reference databases [18]. CheckV provides complementary assessment of completeness and contamination in metagenome-assembled viral genomes [19]. Large resources such as MetaVR and VIRE further expand the searchable viral sequence space and integrate environmental context [5,20]. Yet computational improvement cannot fully compensate for biological underrepresentation.

Evidence: Current viral resources contain millions of previously uncultivated viral genomes and genome-resolved viral populations [5,20]. Their expansion demonstrates the extent to which viral sequence space exceeds traditional reference collections.

Plausibility: Underrepresentation of particular hosts, geographic regions, and ecological systems is likely to reduce the ability of homology-based and machine-learning approaches to classify divergent sequences originating from those systems.

Unknowns: The extent to which reference-database bias alters estimates of African viral diversity, lineage distribution, or public-health relevance remains poorly quantified. It is also unclear how much apparently novel sequence space represents genuinely novel viruses versus divergent members of already recognized viral groups.

The resulting unknown category is heterogeneous. It encompasses: (1) genuinely novel viral lineages; (2) highly divergent members of known viral groups; (3) incomplete viral genomes; (4) endogenous viral elements; (5) cellular sequences misclassified as viral; (6) laboratory or environmental contaminants; (7) sequencing artefacts; (8) environmental viral fragments; and (9) biologically real viruses whose functions remain unknown. These categories must not be treated as equivalent.

3.3. Ecological Asymmetry

African viral diversity cannot be understood from human clinical samples alone. Viruses circulate within complex ecological networks involving wildlife, livestock, vectors, plants, water, soil, and human populations. Global viromics studies demonstrate that viral communities vary systematically with host ecology, geography, and environmental conditions [21,22,23,24]. African ecosystems are particularly dynamic: deforestation, agricultural expansion, urbanization, livestock intensification, wildlife trade, climate variability, and changes in vector distributions modify contact networks between hosts and ecological compartments.

Evidence: Viromics studies demonstrate extensive viral diversity across wildlife, vectors, environmental compartments, and human-associated ecosystems [21,22,23,24].

Plausibility: Rapid changes in land use, host distributions, and human–animal interfaces can alter opportunities for viral circulation and transmission. These processes provide plausible mechanisms through which previously uncharacterized viral populations could move between ecological compartments.

Unknowns: The contribution of specific ecological changes to the emergence, persistence, or spillover of individual African viruses remains incompletely quantified. Ecological plausibility should therefore not be presented as evidence of increased pathogenicity or spillover risk. The fundamental distinction remains: ecological opportunity is not evidence of emergence.

4. The African Viral Reference Paradox

Africa has substantially expanded genomic surveillance capacity, particularly following the COVID-19 pandemic. Yet this expansion has occurred within a global genomic infrastructure whose reference space was constructed over decades and remains unevenly populated [25,26]. We term this the African Viral Reference Paradox: the capacity to generate genomic data can increase faster than the capacity of global databases to contextualize the biological diversity represented by those data.

A laboratory may therefore generate a high-quality sequence from an African sample while classification remains uncertain because its closest evolutionary relatives have not been sampled from the same ecological region—or perhaps anywhere. Reference databases dictate what can be recognized, named, compared, and prioritized, thus heavily biasing scientific attention and public-health interpretation.

The problem is particularly acute for highly divergent RNA viruses [28]. Short reads may lack reliable nucleotide-level matches. Protein-level searches can recover deeper relationships, but these approaches also become uncertain when viral proteins are highly divergent or poorly annotated [1,18,28]. The unknown category must therefore be managed as a structured uncertainty rather than a single taxonomic class. A sequence with no database match should trigger additional characterization, not automatic classification as a novel pathogen. Similarly, a highly divergent genome should not receive elevated public-health priority solely because of its evolutionary novelty.

5. From Viral Dark Matter to Viral Intelligence

The central conceptual transition proposed here is from viral dark matter to viral intelligence. Viral dark matter is an epistemic category: viral diversity that remains undetected, unresolved, or uninterpretable. Viral intelligence is an operational category: contextualized information about viral populations that can support biological interpretation, ecological understanding, and surveillance decisions.

5.1. An Epistemic Ladder for Viral Intelligence

To prevent the common inferential error of treating taxonomic novelty as a surrogate for biological significance, we structure the transition across a nine-level epistemic ladder (Figure 2), grouped into three validation phases:

Phase 1: Bioinformatic & Molecular Validation

Level 1. Detection: Was a viral sequence reproducibly detected?

Level 2. Authentication: Is the signal supported by adequate read depth, assembly quality, negative controls, and reproducibility?

Level 3. Taxonomic Placement: Can the sequence be assigned to an established viral lineage or supported as a previously unrecognized lineage?

Phase 2: Biological & Ecological Characterization

Level 4. Host Association: Is there credible evidence that the viral population is associated with the sampled host?

Level 5. Ecological Persistence: Does the signal recur across individuals, locations, seasons, or time?

Level 6. Biological Activity: Is there evidence of replication, transcription, seroconversion, or tissue association?

Phase 3: Epidemiological & Threat Intelligence

Level 7. Transmission Dynamics: Is there evidence of movement between hosts, vectors, or ecological compartments?

Level 8. Phenotypic Significance: Is there evidence that infection affects host biology or is associated with pathology/disease?

Level 9. Public-Health Threat: Does the available evidence justify intensified surveillance, containment, or clinical intervention?

5.2. Evidence–Plausibility–Unknowns (EPU) Framework

The epistemic ladder can be operationalized through three analytical states:

Evidence: Comprises observations directly supported by data: reproducible sequence detection, genome assembly, taxonomic relationships, host association, recurrence, or experimentally supported biological activity.

Plausibility: Concerns mechanisms that are biologically or ecologically reasonable but not yet demonstrated for a specific viral population. These may include potential host connectivity, effects of land-use change, ecological expansion, or possible transmission pathways.

Unknowns: Encompasses unresolved properties such as replication competence, viability, prevalence, persistence, transmissibility, host range, seroprevalence, and disease causation.

For most metagenome-assembled viral genomes, sequence detection provides limited information about replication, infectivity, transmission, or disease causation. These uncertainties define the exact boundary between raw viral discovery and actionable viral intelligence.

5.3. Novelty Is Not Risk

A highly divergent environmental virus may represent substantial evolutionary novelty but little immediate public-health concern. Conversely, a less novel virus repeatedly detected at a human–animal interface may warrant urgent attention because of its ecological connectivity and changing distribution. Risk prioritization must therefore integrate multiple dimensions: Novelty × Evidence × Host Connectivity × Geographical Trajectory × Public-Health Relevance. This multidimensional approach avoids the “novelty trap,” in which the most evolutionarily unusual sequence automatically receives the highest priority.

6. Methodological Workflows and Emerging Technologies

6.1. Shotgun Metagenomics and Metatranscriptomics

Shotgun metagenomics provides broad access to DNA- and RNA-derived nucleic-acid sequences, whereas metatranscriptomics improves the detection of actively transcribed RNA viruses. Neither approach is universally optimal. Library preparation, nucleic-acid extraction, viral enrichment, depletion strategies, sequencing depth, and specimen type profoundly influence the resulting virome [29,30]. African viral surveillance should avoid imposing a single universal laboratory protocol; instead, programmes should define standardized minimum reporting and quality-control requirements while allowing laboratory workflows to adapt to specimen type and research question. The goal is comparability without sacrificing biological sensitivity.

6.2. Long-Read Sequencing

Long-read sequencing (e.g., Oxford Nanopore, PacBio) improves genome reconstruction and bridges repetitive or complex regions that remain fragmented in short-read datasets. Recent mosquito virome studies illustrate its potential for resolving viral diversity across different sampling scales [31]. However, long reads do not eliminate contamination, misclassification, or biological ambiguity. Their value is maximized when integrated with short-read sequencing, independent validation, and stringent quality control.

6.3. Artificial Intelligence and Structural Homology

Artificial intelligence is transforming viral discovery. As depicted in the integrated computational workflow in Figure 3, large-scale computational models and protein-structure-based prediction algorithms (e.g., AlphaFold, ESMFold) can identify deep evolutionary relationships that escape conventional sequence similarity searches [32,33]. This capability expands viral dark-matter discovery by inferring functional annotations across highly divergent sequence space. But AI introduces a distinct epistemic risk: a prediction is not an observation. Models trained predominantly on existing databases may reproduce the taxonomic, geographical, and ecological biases embedded in those databases. High model complexity does not guarantee reliability for underrepresented viral lineages. African viral intelligence requires: (1) explicit uncertainty estimates; (2) transparent training-data provenance; (3) independent experimental validation; (4) benchmark datasets containing divergent African viral groups; and (5) rigorous separation of model prediction from experimentally supported observation.

6.4. Integrated Computational Workflows

Modern viral discovery combines viral identification, genome-quality assessment, taxonomic classification, and structural protein inference. VirSorter2 and CheckV provide indispensable standard components [18,19]. However, indiscriminately chaining bioinformatic tools can compound false-positive and false-negative errors [34]. The analytical goal is not the largest possible tally of candidate contigs, but a fully validated evidence chain.

7. The Major Methodological Trap: Discovering Noise Instead of Viruses

The rapid expansion of viral discovery creates a critical methodological risk: becoming increasingly efficient at detecting sequences without becoming equally effective at identifying biologically real viruses. Five specific failure modes must be guarded against:

1.Contamination: Environmental specimens contain high nucleic acid backgrounds. Reagents, extraction kits, sequencing libraries, and cross-sample carryover introduce sequences unrelated to the biological specimen. Negative extraction and sequencing controls are obligatory, and low-abundance signals must be evaluated relative to batch controls.

2.Endogenous Viral Elements (EVEs): Viral sequences integrated into host genomes can mimic active viruses. This is highly prevalent in wildlife, where EVEs represent substantial evolutionary memory [35]. The presence of a viral-like sequence in a host-derived sample does not prove active viral infection or replication.

3.Assembly Artefacts: Short-read assemblies can generate chimeric or fragmented contigs. While CheckV assesses completeness [19], high completeness must never be conflated with biological viability.

4.Index Misassignment and Carryover: High-throughput multiplexing can introduce low-level index hopping, generating false-positive rare infection signals. Signal reproducibility across independent biological replicates and sequencing batches is essential.

5.Database Contamination and Annotation Propagation: Reference databases contain misannotated sequences that propagate iteratively through automated pipelines, turning initial misclassifications into consensus errors.

Every candidate African viral dark-matter sequence must therefore receive a formal confidence profile incorporating read support, assembly quality, genome completeness, contamination indicators, taxonomic confidence, host association, recurrence, geographical distribution, biological activity, and orthogonal validation. A sequence must never be classified as an emerging pathogen solely because it is taxonomically novel.

8. From Novel Virus to Risk-Relevant Virus: The VRPS Framework

The public-health utility of viral dark-matter surveillance depends on structured prioritization. Without it, discovery programmes generate vast lists of viral genomes with minimal actionable utility. We propose the **Viral Relevance and Priority Score (VRPS)** as a transparent, standardized decision-support tool. The VRPS is not a deterministic predictive model and does not calculate absolute spillover probability; its objective is to organize heterogeneous evidence and prioritize candidates for targeted follow-up.

8.1. Five-Dimensional Scoring Metric

Each candidate viral population is evaluated from 0 to 4 across five weighted dimensions

Dimension	0	1	2	3	4	Weight
Novelty (N)	Known/near-identical	Minor divergence	Moderately divergent	Highly divergent lineage	Previously unrecognized lineage	15%
Evidence (E)	Sequence signal only	Weak reproducibility	Reproducible sequence/assembly	Evidence of replication, exposure or persistence	Independent biological validation and/or disease association	25%
Host connectivity (H)	Single environmental compartment	Single non-human host	Multiple related hosts	Human–animal or vector interface	Repeated detection across epidemiologically connected hosts/compartments	20%
Geographical trajectory (G)	Single detection	Stable local detection	Recurrent local circulation	Expansion across sites/regions	Consistent spatial expansion or emergence in new ecological settings	20%
Public health relevance (P)	No plausible relevance	Low plausibility	Uncertain relevance	Plausible relevance based on host, biology or exposure	Strong rationale for intensified surveillance	20%

The proposed score is: VRPS = 100 × [(0.15N + 0.25E + 0.20H + 0.20G + 0.20P) / 4] The resulting value ranges from 0 to 100.

The weighting deliberately gives greater influence to evidentiary strength than to taxonomic novelty. These weights are provisional and should be prospectively evaluated against longitudinal surveillance outcomes.

8.2. Operational Action Tiers

VRPS	Priority	Operational interpretation
0–24	Background	Maintain in the Atlas; no immediate escalation
25–49	Monitor	Repeat sampling and improve metadata/evidence
50–74	Investigate	Targeted longitudinal surveillance and orthogonal validation
75–100	Intensify	Multidisciplinary investigation and expanded surveillance

The operational implementation of the VRPS calculation engine and its integrated decision dashboard are detailed in Figure 4.

8.3. Evidence Gates and Safeguards

To prevent score inflation from low-quality data, candidates must pass five strict Evidence Gates before scoring: (1) Authentication Gate: Minimum read depth and negative control verification; (2) Contamination Gate: Zero unresolved reagent or cross-sample contamination; (3) Biological-Context Gate: Complete host, spatial, and ecological metadata; (4) Activity Gate: Explicit distinction between contig presence and active replication; (5) Independence Gate: VRPS escalation above 50 requires orthogonal verification or independent sample re-detection.

8.4. Trajectory and Dynamic Re-Evaluation

Emergence is a longitudinal process. A viral population detected once in an isolated host differs fundamentally from one undergoing spatial expansion or host shifts [21,36,37,38]. The VRPS must be dynamically recalculated as longitudinal data accumulate. A candidate can escalate from Monitor to Investigate without any genomic mutation if its geographical trajectory or host connectivity expands. In anticipatory surveillance, the primary signal of concern is often a change in ecological context rather than a change in viral sequence.

9. From Pathogen Surveillance to Ecosystem Surveillance

Surveillance focused exclusively on human clinical cases captures infection only after spillover has occurred. As structured in the One Health framework in Figure 5, a comprehensive ecosystem surveillance system must systematically sample four interconnected ecological compartments:

1.Human Compartment: Unexplained febrile cohorts, severe acute respiratory or neurological syndromes, occupational sentinel populations (hunters, slaughterhouse workers).

2.Animal Compartment: Domestic livestock, wildlife reservoirs (bats, rodents, non-human primates), animals in trade networks and live markets.

3.Vector Compartment: Mosquitoes, ticks, sandflies, and other hematophagous arthropods.

4.Environmental Compartment: Wastewater, surface water, agricultural runoff, and air sampling at high-contact ecological interfaces [39].

The objective is not blanket, unselective sequencing of all environments, which is neither cost-effective nor informative. Instead, sampling must be strategically targeted at ecological hotspots where land-use changes, biodiversity loss, or altered contact networks create high exposure potential.

10. The African Viral Dark Matter Atlas

We propose the African Viral Dark Matter Atlas (AVDMA) as a continent-wide infrastructure for contextualized viral intelligence. Rather than acting as a static sequence repository, the AVDMA integrates genomic data with ecological, spatial, and temporal metadata.

Every AVDMA entry follows a unified metadata structure: Sequence → Genome Quality → Taxonomy → Host → Location → Environment → Time → Abundance → Recurrence → Ecological Association → Biological Evidence → Risk Indicators. As outlined in Figure 4, the Atlas operates through three integrated layers:

Genome Layer: Assembled genomes, contigs, protein annotations, CheckV quality metrics, evolutionary trees, and AI structural models.

Ecological Layer: Host and vector distributions, land-use mapping, climate variables, seasonality, and agricultural interfaces.

Surveillance Layer: Spatial-temporal tracking, prevalence trends, serology, VRPS scores, and escalation alerts.

This architecture transforms the central inquiry from “What is this sequence?” to “Where does this viral population circulate, in which hosts, under what ecological drivers, and has its risk profile changed?” [5,20]. Addressing the four structural knowledge gaps shown in Figure 6 is crucial to achieving this transition.

Figure 6. The four quadrants of structural knowledge asymmetry in global viral surveillance. Unresolved viral dark matter stems from four interconnected analytical gaps: Quadrant 1 (Sampling Gap: historical focus on severe human clinical cases leading to severe under-sampling of wildlife and environmental reservoirs in Africa), Quadrant 2 (Database Gap: severe under-representation of African-origin viral genomes in public databases like NCBI/GenBank), Quadrant 3 (Diagnostic & Bioinformatic Gap: alignment failure or ambiguous classification of divergent reads due to sparse reference data), and Quadrant 4 (Surveillance & Preparedness Gap: time lag between unobserved transmission in animal reservoirs and human emergence). Addressing these asymmetries is essential to close the preparedness gap.

Figure 6. The four quadrants of structural knowledge asymmetry in global viral surveillance. Unresolved viral dark matter stems from four interconnected analytical gaps: Quadrant 1 (Sampling Gap: historical focus on severe human clinical cases leading to severe under-sampling of wildlife and environmental reservoirs in Africa), Quadrant 2 (Database Gap: severe under-representation of African-origin viral genomes in public databases like NCBI/GenBank), Quadrant 3 (Diagnostic & Bioinformatic Gap: alignment failure or ambiguous classification of divergent reads due to sparse reference data), and Quadrant 4 (Surveillance & Preparedness Gap: time lag between unobserved transmission in animal reservoirs and human emergence). Addressing these asymmetries is essential to close the preparedness gap.

11. Implications for Pandemic Preparedness and African Scientific Sovereignty

The primary justification for an African viral dark-matter programme is anticipatory preparedness. Sequence-based models cannot predict exact pandemic emergence events because spillover depends on complex ecological, host, viral, and human behavioral variables. However, systematic surveillance reduces baseline uncertainty. By characterizing viral populations before spillover occurs, baseline diagnostic assays can be updated, diagnostic blind spots eliminated, and early warning signals captured.

11.1. African Scientific Sovereignty and Governance

Historical models of viral discovery in Africa have frequently suffered from extractive dynamics: biological specimens collected locally were exported for sequencing and publication abroad, leaving local infrastructure underfunded. The AVDMA framework asserts that African institutions must exercise primary leadership over sampling design, sequencing, bioinformatic analysis, data stewardship, and policy interpretation. Local contextual knowledge regarding seasonal human behavior, ecology, and clinical presentation is an indispensable component of scientific accuracy. Data-sharing protocols must balance international open-science standards with national sovereignty, ethical governance, and equitable benefit-sharing.

12. Research Agenda for the Next Decade

To operationalize the framework, we outline five strategic research priorities for the next decade:

1.Establish Representative Baseline Viromes: Conduct systematic, longitudinal virome mapping across contrasting ecosystems (tropical rainforests, savannahs, urban fringes) to define baseline viral diversity [40,41,42,43].

2.Standardize Evidence Frameworks: Enforce rigid epistemic standards distinguishing simple contig detection from proven biological activity and disease causation.

3.Transition to Longitudinal Surveillance: Shift resources from single-point discovery expeditions to sustained, multi-year monitoring of high-risk human–animal interfaces[44].

4.Integrate Multi-Omics and Environmental Intelligence: Combine viral genomic data with remote sensing, land-use cartography, vector ecology, and climate modeling.

5.Build African Analytical Sovereignty: Invest heavily in local computational infrastructure, training African bioinformaticians, structural biologists, and genomic epidemiologists[45,46,47].

13. Conclusion

The primary bottleneck in global pandemic preparedness is not a lack of sequencing throughput, but an over-reliance on static diagnostic targets and pre-defined pathogen lists. African Viral Dark Matter represents a structural knowledge asymmetry born from historical sampling biases, uneven reference coverage, and methodological constraints. By uniting pathogen-agnostic metagenomics, structural AI, One Health ecological integration, and the VRPS framework within an African-led governance model, surveillance can evolve from reactive outbreak response into a proactive, anticipatory intelligence system. The objective is not to find every virus on the continent, but to systematically illuminate the biological unknowns that constrain global health security.