1. Working with Agentic AI: Bioinformatics Annotator

How human skills and agentic skills are dynamically integrated and coordinated across bioinformatics annotation responsibilities

1.1. Bioinformatics Annotator as an Occupation

A bioinformatics annotator assigns biological meaning, functional classifications, and structured metadata to raw molecular data, genomic sequences, and biological literature. This occupation connects molecular biology, biochemistry, and data science by transforming uncharacterized sequences, structural models, and unstructured scientific texts into standardized, computable biological knowledge. To ensure that these annotations accurately reflect current scientific consensus and support reliable downstream computational research, the annotator must understand how biological nomenclature, standardized ontologies, experimental evidence codes, and curation guidelines collectively determine the validity and interoperability of biological databases.

Typical responsibilities of a bioinformatics annotator include:

• genomic and sequence feature annotation;
• functional and ontology annotation;
• variant classification and phenotypic annotation;
• literature curation and evidence synthesis;
• annotation quality assurance and conflict resolution; and
• training data generation and AI model evaluation.

These responsibilities are fulfilled through distinct tasks and subtasks, each requiring one or more actions with specific skill requirements. Human skills and agentic skills both contribute to these responsibilities, but their levels and forms of involvement differ according to the responsibility, task, action, data, biological context, technical environment, and expected outcome.

1.2. Human Skills and Agentic Skills in Bioinformatics Annotation

Bioinformatics annotation increasingly integrates both human skills and agentic skills.

Human skills are reusable capabilities possessed and developed by bioinformatics annotators. They include biological domain expertise, ontology reasoning, critical literature evaluation, evidence code interpretation, nomenclature governance, structural biology intuition, and scientific judgment.

Agentic skills are reusable capabilities available to AI agents through their underlying models, contextual information, instructions, constraints, tools, and computational resources. In bioinformatics annotation, these skills include natural language processing for literature mining, automated sequence alignment, ontology mapping, database querying, pattern recognition in genomic sequences, structural prediction execution, and draft annotation generation.

Both skill types contribute to the same occupational responsibilities, but their involvement varies by action. Understanding the precise biological function of a newly discovered protein relies primarily on human knowledge of metabolic pathways and experimental evidence strength; an AI agent may then use agentic skills to mine thousands of research abstracts, map protein domains to standardized ontology terms, and retrieve homologous sequences from public repositories. In variant classification, the annotator first determines the appropriate classification framework and weighs conflicting clinical or phenotypic evidence, while an AI agent subsequently aggregates population frequencies, predicts structural disruptions, and retrieves related case reports. Across these actions, human and agentic skills are activated sequentially or iteratively based on task requirements, with the human professional maintaining oversight and final judgment.

The relationship between these skills is not a fixed division where certain responsibilities permanently belong to humans and others to AI. Instead, human and agentic involvement shifts across tasks and between different actions within the same task.

1.3. Dynamic Human–Agentic Skill Integration and Collaboration in Bioinformatics Annotation

Bioinformatics annotation functions can be understood as an occupation where responsibilities are fulfilled through tasks and subtasks, tasks and subtasks are executed through actions, and actions require appropriate configurations of human and agentic skills.

An occupation provides the overall professional context. For bioinformatics annotation, this context encompasses characterizing molecular sequences, assigning standardized biological functions, curating scientific literature, and maintaining the integrity of computational knowledgebases.

A responsibility represents an expected area of professional work or an outcome that the bioinformatics annotator accomplishes or maintains. Responsibilities such as sequence annotation, ontology mapping, literature curation, and quality assurance each contain multiple tasks.

A task is a context-dependent unit of work used to fulfill part or all of a responsibility. When a task is too broad or complex for reliable execution, it is decomposed into smaller subtasks.

Tasks and subtasks are executed through one or more actions. Actions represent reusable and configurable operations such as aligning a novel transcript, mapping a gene to a controlled vocabulary, extracting a protein-protein interaction from a text, classifying a genetic variant, or updating a database record.

Each action has skill requirements. These requirements describe the capabilities needed for effective and reliable execution. Depending on the action and its context, human skills, agentic skills, or an integration of both can satisfy these requirements.

The skills available for bioinformatics annotation reside in a skill space containing both human skills and agentic skills. Relevant skills in this space can serve as candidate skills for executing the actions required by a particular task. A human skill in evidence code interpretation, for example, can serve as a candidate for determining whether a high-throughput screening paper provides sufficient proof for a specific gene-disease association, while an agentic skill in named entity recognition can serve as a candidate for extracting the relevant genes and diseases from that same paper.

Candidate skills are evaluated relative to the requirements of the action and its occupational and task context. Appropriate skills are then selected and organized into a skill configuration for execution. Depending on these requirements, a configuration may rely mainly on human skills, mainly on agentic skills, or on an integrated arrangement of both.

1.3.1. Skills Competition and Skills Collaboration

Two important relationships can occur between human skills and agentic skills when constructing a skill configuration: skills competition and skills collaboration.

Skills competition occurs when both a human skill and an agentic skill can satisfy the same or similar requirements for an action, but simultaneous application of both is unnecessary. Under a particular occupational and task context, one skill type can provide a better fit for the action than the other.

For example, both a bioinformatics annotator and an AI agent can assign standard functional terms to a well-characterized protein family based on sequence homology. When the sequence similarity is extremely high and the functional definitions are unambiguous, the agentic skill provides fast and scalable execution. When the annotation requires distinguishing between a direct catalytic function and an indirect regulatory role based on nuanced structural differences or conflicting literature, the annotator’s human skills provide the necessary biological reasoning.

The purpose of skills competition is not to establish whether human or agentic skills are generally superior. Rather, it identifies which available skill provides the better fit for a particular action under particular conditions.

Skills collaboration occurs when one skill type alone cannot effectively satisfy the requirements of a task or its actions, or when combining complementary human and agentic skills produces stronger task performance.

Collaboration does not require human and agentic skills to operate simultaneously. Instead, they are activated at different stages of action execution in a sequential, iterative, or recursive manner according to evolving task requirements.

For example, an annotator first identifies a novel genomic region suspected to contain a complex non-coding RNA cluster. An AI agent then executes multiple ab initio prediction algorithms, aligns the region against diverse species, and extracts relevant structural motifs. The annotator subsequently evaluates these computational predictions against known biological constraints and formulates a consensus structural model. Based on the annotator’s direction, the agent next searches specialized databases to verify the conservation of the proposed secondary structure. Finally, the annotator evaluates the resulting evidence and assigns the final standardized feature annotations. Throughout this sequence, human and agentic skills are activated iteratively as task requirements evolve, while the annotator directs the investigation and evaluates its outcomes.

This dynamic relationship allows bioinformatics annotators to leverage the complementary strengths of both skill types rather than relying on a permanent allocation of work.

1.3.2. Skill Configuration and Task Execution

A skill configuration specifies which human skills and agentic skills are used to execute the actions associated with a task or subtask and how those skills are coordinated.

The configuration depends on factors such as action requirements, data volume, biological ambiguity, evidence strength, expected reliability, available databases, information standardization levels, downstream research consequences, and intermediate results.

For one action, an agentic skill may independently satisfy the requirements. For another, a human skill may provide the better fit. A complex task may require several configurations across multiple actions, with human and agentic skills repeatedly interacting as the task progresses.

The configuration is therefore dynamic rather than permanent. It may change when the action changes, new biological evidence is published, an agentic result fails to satisfy expected curation guidelines, or human interpretation identifies a new nomenclature conflict.

Throughout this process, the bioinformatics annotator remains the professional responsible for directing and controlling the work associated with the occupation and its responsibilities. Agentic AI provides powerful complementary capabilities that expand what the professional can mine, align, map, classify, verify, and document.

1.3.3. Task Performance Evaluation and Skill Space Adaptation

The performance produced through a selected skill configuration is evaluated against predefined evaluation metrics. These metrics determine whether the human skills, agentic skills, or their integrated configuration are sufficiently capable of performing the relevant task and its required actions.

Evaluation criteria depend on the work being performed. In bioinformatics annotation, they include biological accuracy, ontology compliance, evidence code correctness, literature extraction precision, database consistency, nomenclature standardization, and other task-specific expectations.

When a selected skill or skill configuration produces performance above the defined threshold, it passes the performance evaluation and is retained for future applications under appropriate conditions. Repeated successful use provides additional evidence about where that skill or configuration performs effectively.

When performance falls below the required threshold, the corresponding skill or configuration is not reused without modification. The relevant human or agentic skills may be refined or reconfigured, existing skills can be combined differently, or new human and agentic skills may be developed or constructed from scratch when the existing skill space lacks adequate capability.

For example, an AI agent first extracts gene-disease associations from a new batch of open-access literature, but repeatedly misinterprets negative results (e.g., “Gene X is not associated with Disease Y”) as positive associations. Task performance evaluation identifies this extraction error as falling below the required threshold. The agentic skill is then refined through improved natural language processing instructions, specific negation-detection logic, or fine-tuning on a specialized biological corpus. The revised skill is evaluated again, and once it satisfies the required performance criteria, it is retained for appropriate future literature curation tasks. Human skills follow a corresponding development process: when an annotator encounters a newly revised international nomenclature committee guideline for naming complex immune receptors, the relevant capability is developed through studying the documentation and practicing on edge cases before its performance is evaluated through subsequent database updates.

Refined and newly constructed skills that satisfy the required performance criteria are added to the skill space. In this way, task performance evaluation provides feedback not only for completing current work but also for continuously optimizing and expanding the available skill space.

The skill space is not constructed as an isolated collection of occupation-specific representations. Many human and agentic skills used in bioinformatics annotation are reusable across occupations. Literature mining, database querying, ontology mapping, statistical reasoning, pattern recognition, summarization, and documentation contribute to work in clinical data analysis, pharmacovigilance, biomedical text mining, data science, and many other fields.

Reusable skills can therefore be represented once and configured differently according to the responsibilities, tasks, actions, and contexts of different occupations. This allows the skill space to serve a growing range of occupations without unnecessarily duplicating or exploding the number of skill representations. Occupational specialization is introduced through task context, action requirements, parameters, biological data sources, tools, constraints, and skill configurations rather than by creating a completely separate skill representation whenever the same underlying capability appears in another occupation.

The resulting process is adaptive: occupational responsibility → task and subtask → actions → skill requirements → candidate human and agentic skills → skill configuration → task execution → task performance evaluation → skill refinement, reconfiguration, or construction → optimized and expanded skill space

This cycle allows human and agentic capabilities to evolve alongside the requirements of bioinformatics annotation work.

1.4. Essential Responsibilities of Bioinformatics Annotator

1.4.1. Genomic and Sequence Feature Annotation

Genomic and sequence feature annotation identifies and characterizes structural elements within DNA and RNA sequences, such as genes, exons, promoters, regulatory regions, and non-coding RNAs. This responsibility includes executing prediction pipelines, comparing homology across species, resolving overlapping transcripts, identifying pseudogenes, and assigning standardized feature coordinates to genomic assemblies.

Agentic skills contribute extensively to automated ab initio gene prediction, homology-based sequence alignment, repeat masking, and draft coordinate generation. Human skills remain essential for resolving complex alternative splicing events, validating the biological plausibility of predicted open reading frames, distinguishing functional genes from transcriptional noise, and deciding how to represent ambiguous structural features in the database.

For example, a newly sequenced microbial genome requires comprehensive structural annotation. An AI agent executes multiple prediction algorithms, aligns the sequences against public protein repositories, and generates a draft set of gene coordinates. The bioinformatics annotator reviews the draft and identifies a region where two predicted genes overlap on opposite strands, a common artifact in automated pipelines. The annotator examines the raw sequencing read coverage and ribosomal binding site motifs to determine the true biological boundaries. Based on the annotator’s corrections, the agent updates the genomic coordinates and formats the final output for database submission.

1.4.2. Functional and Ontology Annotation

Functional and ontology annotation assigns standardized biological meaning to genes and proteins using controlled vocabularies, such as molecular function, biological process, and cellular component ontologies. This responsibility includes mapping protein domains to functional terms, evaluating experimental evidence, assigning appropriate evidence codes, and ensuring that annotations remain consistent with evolving international ontology standards.

Agentic skills are well suited to text-mining abstracts for keyword associations, mapping structural domains to functional databases, and proposing initial ontology terms based on sequence homology. Human skills contribute to evaluating the actual strength of the experimental evidence, distinguishing direct molecular functions from indirect downstream phenotypic effects, and applying strict evidence codes that dictate how computational models may use the annotation downstream.

For example, an AI agent processes a newly characterized plant enzyme and proposes a specific metabolic pathway annotation based on sequence similarity to a known bacterial enzyme. The bioinformatics annotator reviews the proposal and notes that the plant enzyme lacks a critical co-factor binding site identified in the structural literature. The annotator rejects the automated pathway assignment, manually assigns a more general catalytic function term, and attaches an evidence code indicating that the specific pathway role remains unverified. The agent then logs this human-verified exception to prevent similar automated misannotations in related plant families.

1.4.3. Variant Classification and Phenotypic Annotation

Variant classification and phenotypic annotation evaluates genetic variations and assigns clinical, agricultural, or functional significance based on population data, structural impact, and experimental evidence. This responsibility includes aggregating allele frequencies, interpreting standardized classification frameworks, weighing conflicting literature, and linking specific variants to observable traits or disease states.

Agentic skills retrieve population frequencies from diverse genomic databases, predict structural disruptions caused by amino acid substitutions, and aggregate related case reports from medical or agricultural literature. Human skills evaluate the clinical or phenotypic significance of the variant, apply standardized classification guidelines (e.g., pathogenic, benign, trait-associated), resolve conflicting evidence between computational predictions and functional assays, and determine the final classification.

For example, an AI agent flags a novel missense variant in a crop resistance gene, predicting severe structural disruption and retrieving three isolated field reports suggesting susceptibility. The bioinformatics annotator reviews the structural model and the field reports, recognizing that the reports lack proper control groups and the structural prediction ignores a known compensatory mechanism in this specific plant family. The annotator classifies the variant as “uncertain significance” rather than “deleterious,” and instructs the agent to draft a summary of the conflicting evidence for the database entry, ensuring downstream breeders are appropriately cautioned.

1.4.4. Literature Curation and Evidence Synthesis

Literature curation and evidence synthesis extracts specific biological relationships—such as protein-protein interactions, gene-disease associations, or metabolic pathways—from published scientific texts. This responsibility involves reading complex methodologies, verifying the biological context (e.g., in vivo vs. in vitro, species specificity), identifying false positives from high-throughput screening papers, and structuring the extracted data for database ingestion.

Agentic skills perform named entity recognition, relationship extraction, full-text parsing, and initial summarization of findings across thousands of papers. Human skills verify the biological context, identify methodological flaws that invalidate the reported interactions, resolve ambiguous author claims, and ensure that the extracted relationships accurately reflect the physical reality described in the paper.

For example, an AI agent scans a high-throughput proteomics paper and extracts hundreds of potential protein-protein interactions, automatically linking them to the relevant gene identifiers. The bioinformatics annotator reviews a sample of these extractions and realizes that the paper’s methodology frequently captures indirect co-complex members rather than direct physical interactors. The annotator adjusts the curation parameters, instructing the agent to apply a specific interaction evidence code that denotes “physical association within a complex” rather than “direct binding,” thereby preserving the data’s utility while preventing misleading downstream network models.

1.4.5. Annotation Quality Assurance and Conflict Resolution

Annotation quality assurance and conflict resolution maintains database integrity by reviewing automated annotations, resolving contradictions between different data sources, and updating records as scientific consensus evolves. This responsibility includes continuous database scanning, investigating systematic errors, arbitrating conflicts between computational predictions and experimental data, and updating curation guidelines.

Agentic skills contribute to continuous database scanning for logical inconsistencies, cross-referencing external repositories, flagging outdated entries, and generating discrepancy reports. Human skills investigate the root cause of systematic annotation errors, arbitrate conflicts based on the hierarchy of biological evidence, and revise the underlying curation rules to prevent future inconsistencies.

For example, an automated quality control script alerts the team that a specific gene family is annotated with two mutually exclusive molecular functions across different database releases. An AI agent traces the origin of the annotations, identifying that one set derives from an older text-mining pipeline while the other comes from recent manual curation. The bioinformatics annotator reviews the evidence, confirms that the recent manual curation is correct based on new crystallography data, and authorizes the agent to systematically deprecate the outdated text-mined annotations across all related species in the database.

1.4.6. Training Data Generation and AI Model Evaluation

Training data generation and AI model evaluation creates high-quality, manually verified “gold-standard” datasets used to train, benchmark, and improve new bioinformatics machine learning tools. This responsibility involves defining edge cases, performing rigorous manual verification, establishing the ground-truth biological reality, and calculating initial performance metrics for novel algorithms.

Agentic skills generate baseline predictions, format complex data structures, calculate initial performance metrics, and identify areas of high model uncertainty. Human skills perform the rigorous manual verification required to establish ground truth, define the biological edge cases that challenge current algorithms, and evaluate whether the AI model is learning genuine biological principles or merely memorizing dataset artifacts.

For example, a research team is developing a new deep learning model to predict the subcellular localization of uncharacterized proteins. The bioinformatics annotator defines a diverse set of edge cases, including proteins that shuttle between compartments or require specific post-translational modifications for localization. An AI agent formats these sequences, runs the baseline models, and highlights the proteins where the new model’s confidence is lowest. The annotator manually verifies the true localization of these difficult cases using primary literature, creating a refined benchmark dataset that forces the next iteration of the AI model to improve its biological reasoning.

1.5. Changing Human and Agentic Involvement Across Bioinformatics Annotator Responsibilities

Human skills and agentic skills are involved throughout bioinformatics annotation, but their relative involvement differs across responsibilities and changes during task execution.

These involvement levels are not permanent allocations. A single responsibility may contain actions where human and agentic skills compete as alternative capabilities and other actions where they collaborate sequentially, iteratively, or recursively.

The practical objective is to construct skill configurations that use the most suitable available capabilities for each part of the work and revise those configurations when task performance evaluation shows that another arrangement performs more effectively or reliably.

1.6. Working as a Bioinformatics Annotator with Agentic AI

Agentic AI is becoming an increasingly powerful component of bioinformatics annotation. Bioinformatics annotators should therefore adopt and integrate AI agents as complementary participants in professional workflows rather than treating them only as occasional tools for text mining, sequence alignment, or database formatting.

This shift does not reduce the importance of deep biological expertise and curation rigor. Human professionals remain responsible for the occupation and its corresponding responsibilities. They understand molecular mechanisms, define meaningful biological relationships, interpret experimental evidence, determine appropriate ontology mappings, evaluate variant significance, communicate curation standards, control consequential database updates, evaluate task performance, and decide how human and agentic skills are configured and improved.

At the same time, agentic skills greatly expand the volume and variety of biological literature and sequence data that can be processed. AI agents retrieve and organize massive volumes of scientific texts, generate and test structural alignments, map heterogeneous data to universal ontologies, calculate population metrics, extract complex relationships, draft curation summaries, and synthesize extensive biological evidence at a scale difficult to achieve through direct human effort alone.

Strong bioinformatics annotation practice is therefore not based on maximizing human work or agentic work independently. It is based on finding appropriate configurations of human skills and agentic skills for the responsibilities, tasks, subtasks, and actions being performed.

Where a human skill and an agentic skill can independently perform the same action, skills competition identifies which is better suited to the current context. Where work requires complementary capabilities, skills collaboration results in human and agentic skills being activated sequentially, iteratively, or recursively to achieve stronger task performance than either skill type achieves alone.

Task performance evaluation then determines whether selected skills and configurations satisfy predefined expectations. Successful skills are retained and reused, while inadequate skills are refined, reconfigured, or newly developed. The resulting capabilities return to an increasingly useful skill space that supports bioinformatics annotation and, where capabilities are reusable, other occupations as well.

For a bioinformatics annotator, working effectively with agentic AI means treating AI as a powerful complementary capability integrated into everyday professional routines while maintaining human control of occupational responsibilities. Through appropriate skill selection, competition, collaboration, evaluation, and continuous improvement, human professionals and agentic AI work together to achieve bioinformatics annotation that is more efficient, scalable, adaptable, timely, standardized, and biologically profound.