# Biosafety Risk Assessment for BSL-3 Labs: A Step-by-Step Guide BSL-3 stands for [biosafety level](/knowledge/diagnostics/emerging-tech/biosafety-levels-1-4-a-comparative-framework-for-diagnostic-laboratory-design) 3, the containment tier for work with agents that can cause serious or potentially lethal disease through inhalation, such as Mycobacterium tuberculosis, Yersinia pestis, or respiratory viruses handled at high concentration. A BSL-3 risk assessment is the documented process of identifying every plausible route by which those agents could reach a worker, the public, or the environment, rating each route for likelihood and consequence, and selecting controls that bring the remaining risk to a level the institution accepts in writing. This guide walks through a biosafety risk assessment for BSL-3 labs in the order working laboratories actually use it. The method achieves four things. It produces an agent and procedure inventory that matches reality on the bench. It verifies that the room, its airflow, and its filters still perform as designed. It converts judgment into a scored, comparable risk register. It leaves behind a document that survives an audit and an incident investigation. Use a full assessment when a suite is commissioned, when a new agent or a new aerosol-generating procedure is introduced, after renovation or HVAC work, after any exposure or near miss, and on a fixed annual cycle. Plan on roughly 20 to 30 hours of hands-on work for a first full pass through a small suite, spread over 6 to 10 weeks of elapsed time while engineering checks are scheduled. A yearly update usually needs a few hours of team time plus a walkthrough. ## Principle: How a Biosafety Risk Assessment Works A biosafety risk assessment estimates risk as the product of likelihood and consequence: how often a given exposure could occur, and how bad it would be if it did. The output is a ranked list of hazards, a set of controls matched to each hazard, and a written statement of the risk that remains after those controls are in place. The assessment sits on top of a fixed engineering baseline that defines BSL-3 containment: - **Directional airflow.** Air moves from cleaner adjacent spaces into the laboratory, so any leak at the boundary carries contamination inward, never outward. - **Negative pressure.** The room sits below the pressure of its surroundings, which is what makes the inward direction real rather than aspirational. Gauges at the entry display it continuously. - **HEPA filtration.** Exhaust air passes through high-efficiency particulate air (HEPA) filters, which by definition trap at least 99.97 percent of particles at the 0.3 micrometer size that is hardest to catch. Infectious material travels on dust and droplet nuclei in and above that size range. - **Anteroom.** An intermediate change room between the corridor and the laboratory, with self-closing doors, where gowning and doffing happen. - **Primary containment devices.** Class II biological safety cabinets, ventilated cabinets whose airflow and filters protect the worker, the sample, and the room simultaneously, plus a pass-through autoclave, a steam sterilizer loaded from inside the laboratory and unloaded outside it. Over that baseline, the assessment applies the hierarchy of controls, the ordering principle that ranks responses from most to least protective: elimination of the hazard, substitution of a safer agent or method, engineering controls, administrative controls such as SOPs and training, and personal protective equipment (PPE) last. Two properties of the exercise deserve emphasis. First, it is iterative. Every control selected in step 8 changes the likelihood scores assigned in step 7, so the team cycles through scoring and control selection until the register stabilizes. Second, it is a living document. The assessment describes a specific room, a specific agent list, and a specific staff at a specific time, and it loses accuracy as any of those change. ## Materials and Records to Assemble Before You Start Gather the items below before the team meets. The assessment moves at the speed of its records, and a missing filter test report costs a week of scheduling. | Item | Specification or working concentration | Role in the assessment | |--|--|--| | Agent and procedure inventory | Current agent list with stock concentrations and SOP revision dates | Raw material for hazard identification (steps 2 and 3) | | Facility and HVAC records | Floor plans, commissioning reports, filter change log, gauge calibration records | Baseline for the engineering review (step 4) | | Differential pressure gauges | Continuous readout at the room boundary, calibrated on a set schedule | Confirms sustained negative pressure | | Smoke pencil or fog source | Non-irritating visible aerosol | Directional airflow checks at doors and penetrations | | HEPA filter test records | Results of in-place leak tests for each installed filter | Evidence that filtration matches design | | Cabinet certification records | Annual certification for each biological safety cabinet | Confirms primary containment devices | | Sodium hypochlorite (household bleach) | Diluted 1:10 to about 0.5 percent available chlorine (roughly 5,000 ppm), prepared fresh the day of use | Surface disinfection during walkthroughs and spill drills | | Ethanol or isopropanol, 70 percent | Ready-to-use dilution | Wiping items and surfaces that chlorine would corrode | | Biological indicators | Geobacillus stearothermophilus spore preparations, 10^6 (one million) spores per unit | Autoclave validation challenge | | Respiratory protection | Fit-tested N95 filtering facepiece respirators or powered air-purifying respirators (PAPRs), per the agent profile | Last-line control for aerosol tasks | | Risk matrix template | Five-by-five likelihood and consequence grid with written band definitions | Consistent scoring across the team | | Incident and exposure logs | Every event since the previous assessment | Feeds the feedback loop in step 12 | **Safety note.** Prepare chlorine dilutions fresh, label them with date and time, and never mix them with acids or ammonia, because the combination releases toxic chlorine gas. Alcohol-based disinfectants are flammable, so keep them away from open flame and hot autoclave surfaces. Respirators protect only when fit-tested to the individual face, so schedule fit-testing before the walkthrough, not after it. ## The BSL-3 Risk Assessment Procedure, Step by Step ![Flow of twelve steps from scope and team through agent profiles, aerosol mapping, scoring and controls to residual risk and the review clock.](/article-images/illustrations/bsl-3-risk-assessment-assessment-steps.webp) The sequence below reflects how experienced teams run a bsl-3 risk assessment step by step. Each step lists the action and the reason it matters. 1. **Set the scope and convene the team.** Write down what the assessment covers: which rooms, which agents, which procedures, and which support activities such as media preparation, waste handling, equipment maintenance, and animal care where applicable. Convene a standing team for the life of the assessment: the biosafety officer, who owns the program, the laboratory director or supervisor, a senior technologist who performs the work daily, the facility or HVAC engineer, an occupational health representative, and animal care staff where relevant. **Reason.** Assessments written by one person away from the bench miss the hazards that matter. The technologist knows which centrifuge cups stick and which gauge reads wrong. The engineer knows which damper sticks in winter. 2. **Characterize every agent in the inventory.** For each agent, record disease severity in humans, the primary exposure route (inhalation for BSL-3 agents), infectious dose where established, stability in the environment and on surfaces, availability of licensed vaccines and effective treatment, and whether the agent is indigenous to the region or exotic. Pull this from agent summary statements and standard reference texts rather than memory. **Reason.** The agent profile drives every later decision. An agent with high inhalation infectivity, no vaccine, and limited treatment pushes engineering and practice controls up the scale. An agent with effective prophylaxis changes the medical surveillance plan instead. 3. **Map procedures and flag aerosol-generating steps.** Walk each standard operating procedure (SOP) and mark every step that can create an aerosol, meaning airborne particles carrying infectious material. Typical flags: centrifugation and the opening of rotor or cup lids, vortexing, sonication, blending, grinding, pipetting and mixing, sharps handling, spill cleanup, and, in animal work, cage changing and bedding dumping. Record where in the room each step happens. **Reason.** Location and timing change exposure. Modeling of a working BSL-3 laboratory showed that infection risk from an airborne release differed by more than 20-fold between locations in the same room, with the highest exposure at the start of a leak [1]. A step performed two meters from the exhaust grille is not the same step performed in a corner recirculation zone. 4. **Inspect engineering controls and the room envelope.** Verify each element of the containment baseline from the principle section: directional airflow, sustained negative pressure on calibrated gauges, a functional anteroom, sealed wall and ceiling penetrations, HEPA-filtered exhaust, certified biological safety cabinets, and a working pass-through autoclave. Published engineering methodology exists for calculating how much reverse airflow a BSL-3 or airborne infection isolation (AII) room, the hospital room type built on the same inward-flow principle, can tolerate before containment is compromised, and facilities with unusual geometry benefit from running that calculation for their own room [2]. **Reason.** Engineering controls protect everyone in the building, including people who never enter the lab, and they fail quietly. A damper that sticks or a gasket that shrinks gives no warning except on paper. 5. **Review work practices and PPE.** Compare each SOP against observed behavior. Check gowning and doffing sequences, glove changes between tasks, sharps handling, surface decontamination at the end of each session, and decontamination of equipment before it leaves the room for maintenance. Check the respiratory protection program: fit-testing records, cartridge change schedules, and PAPR battery and hood checks where those are used. **Reason.** Practice drift is where containment erodes first, and PPE is the last line of defense in the hierarchy, not the first. A respirator program that looks fine on paper fails quietly when fit-testing lapses. **Safety note.** Prepare disinfectant dilutions fresh on the day of use, and never combine chlorine-based products with acids or ammonia. 6. **Audit training, competency, and medical surveillance.** Review the microbiology biosafety training record for every person with room access, including support staff. Training should cover agent hazards and exposure routes, the specific SOPs each person performs, spill and exposure response, and incident reporting, with competency confirmed by direct observation rather than a signed attendance sheet. Confirm enrollment in medical surveillance, baseline serum where offered, and vaccination where a licensed vaccine exists for the agent in use. **Reason.** Documented competency converts paper controls into behavior. Medical surveillance shortens the gap between an unrecognized exposure and effective treatment, which matters most for agents with a treatment window. 7. **Score each hazard on a risk matrix.** Use a five-by-five matrix that multiplies likelihood by consequence, with written definitions for each band, for example low, moderate, and high. Score every hazard line carried forward from steps 2 through 6. Score scenarios, not abstractions: "aerosol release while opening a rotor outside the cabinet" scores differently from "agent handled at all." **Reason.** A consistent scale lets the team rank hazards, compare years, and defend priorities to administrators. Treat the score as a communication tool, not a measurement. 8. **Select controls using the hierarchy.** For each hazard above the acceptable band, select controls in this order: elimination (remove the step), substitution (a safer method, strain, or closed system), engineering controls (cabinet work, sealed rotors and safety cups, interlocks, filtration), administrative controls (SOP changes, access limits, scheduling), and PPE last. **Reason.** Controls higher in the hierarchy protect everyone, including the person who skips a step on a busy day. PPE protects one person, only when worn correctly, and only when it fits. 9. **Quantify the highest-consequence scenarios.** For scenarios whose consequence band is severe, such as a large aerosol release or a sustained equipment failure, add quantitative tools. The Wells-Riley model estimates infection probability from the number of infectious quanta, the model's unit of airborne infectious dose, inhaled over the exposure period. Computational fluid dynamics (CFD) simulates airflow and particle transport through the actual room geometry. In one published BSL-3 study combining field measurement with CFD and Wells-Riley modeling, potential exposure per minute was highest at the start of a leak and settled near 1 percent per minute after 4 minutes, which is the quantitative basis for rapid exit and targeted disinfection procedures [1]. **Reason.** Numbers move engineers and administrators in a way adjectives do not, and the same modeling identifies which corners of the room deserve targeted disinfection after a spill. 10. **Walk the environmental and emergency pathways.** Trace every stream that leaves the room: liquid waste and effluent decontamination, solid waste and its route to the autoclave, sharps, and equipment leaving for repair. Then walk the emergency scenarios: HVAC failure, power loss, fire alarm activation, flood, and structural events. Frameworks developed for high-level biosafety laboratories organize these environmental risk points and the matching emergency measures into a managed system with defined responsibilities [3]. **Reason.** Failures at the environmental boundary expose the public rather than the worker, and they are the failures that neighbors, regulators, and the press ask about first. 11. **Document residual risk and obtain approval.** Residual risk is what remains after controls are applied. Write it down for every hazard line, with the reasoning, the controls selected, and the verification method for each control. Obtain signatures from the biosafety officer and the institutional biosafety committee (IBC), the institutional body that reviews and approves work with biohazardous agents. **Reason.** Signatures create accountability and turn the assessment into the reference document for audits, training, and incident investigations. An unsigned assessment is a draft. 12. **Train, verify, and set the review clock.** Translate every selected control into a specific SOP line item, train against it, and schedule verification: periodic airflow checks, filter testing after any change, autoclave biological indicators on a defined cycle, and behavior audits. Set the next review date inside the document itself. **Reason.** An assessment that does not change SOPs, training, and the verification calendar is a document nobody reads. The review date is what keeps it alive. ### BSL-3 risk assessment tips that prevent rework - Write the assessment around tasks, not rooms. A room gets one score, a task list gets many, and the task list is what staff can act on. - Mark the exact SOP line where aerosols can occur, with the step number, so training and audits point at the same words. - Photograph gauge readings, smoke tests, and filter labels during the walkthrough. Pictures settle arguments later. - Put a daily bench technologist in the scoring session, not just in the cc line. - Define the matrix bands in writing before scoring starts, so the team argues about hazards rather than about words. ## Verification Controls, Expected Results, and How to Read Them An assessment is only as good as the checks that confirm its assumptions. Commissioning, the documented testing that proves installed systems perform to their design, establishes the baseline. The checks below re-confirm it over time. **Directional airflow smoke check.** Release a thin stream of visible smoke at the open anteroom door and at each wall penetration. Expected result: steady inward drift, no outward puffs. Reading: any outward movement fails the check, and containment-dependent work stops until the balance is restored. **Differential pressure monitoring.** Gauges at the entry display the pressure difference between the room and the adjacent space, continuously. Expected result: a stable reading showing the room below ambient, with the alarm functional. Reading: trends matter more than snapshots. A slow drift toward zero usually means filter loading or damper wear, not a sensor fault. **HEPA in-place leak testing.** In-place testing checks the installed filter together with its housing and seals, which is where leaks actually occur. The published record shows why this matters. In a BSL-3 system challenged with aerosolized Serratia marcescens, six of eight first-stage filters and the single second-stage filter left no detectable bacteria downstream, with measured efficiencies above 99.9999 percent. Two first-stage filters measured lower, at 99.9825 and 99.9906 percent, and the fault was traced to a degraded cartridge and an unsealed mounting bracket. After the cartridge was replaced and the bracket sealed, no bacteria were detected downstream anywhere in the system [4]. Reading: treat any decline from baseline as a mounting or seal problem first, then a media problem, and keep every test report so the trend stays visible. **Biological safety cabinet certification.** Expected result: a current certification sticker, face velocity (air speed at the open face) within the certifier's criteria, and no alarms. Reading: an out-of-certification cabinet is a failed control. Work stops until it passes. **Autoclave biological indicators.** Run spore indicators containing Geobacillus stearothermophilus, the heat-resistant spore-forming bacterium used as the standard sterilization challenge organism, with each validation load. Expected result: no growth in the test vial, growth in the unprocessed positive control. Reading: growth in the test vial fails the load. Correct loading and cycle parameters, then repeat with a fresh indicator before releasing the cycle. **Behavior and PPE audits.** Expected result: gowning sequence, glove changes, and respirator use match the SOP on every observation. Reading: repeated gaps in the same step point to a system problem, such as sizing, stockouts, or workload, before they point to a discipline problem. **Safety note.** Space decontamination after a major spill or before major maintenance uses vaporized hydrogen peroxide or formaldehyde. Both demand a sealed room, trained operators, and validated aeration before anyone reenters. ## Troubleshooting Guide | Symptom | Likely cause | Fix | |--|--|--| | Pressure gauge reads near zero or swings with door movement | Damper failure, sensor drift, or a door propped open during readings | Recalibrate the sensor, exercise dampers through full travel, enforce door discipline, then re-commission the room | | Smoke drifts outward at the anteroom door | Airflow imbalance or heavily loaded filters | Rebalance supply and exhaust, inspect filters, confirm interlocked doors close fully on their own | | In-place filter test finds downstream leakage | Gasket leak, unsealed mounting bracket, or damaged media | Reseat and reseal the mounting first, replace the cartridge if the leak persists, then retest before releasing the system | | Biological indicator grows after a normal autoclave cycle | Overloaded chamber, wrong cycle parameters, or wet steam | Correct load configuration and parameters, run a fresh indicator, and repeat until the load passes | | Staff report stale air or odor pockets near equipment | Recirculation zones around large equipment, weak air exchange above bench height | Space out or relocate equipment, verify exhaust balance, and target disinfection at the affected zones | | Every hazard scores low year after year | Anchoring on previous scores and missing frontline input | Re-score from written scenario descriptions with a bench technologist in the room | | Audits keep finding the same PPE gaps | Fit range problems, stockouts, or workload pressure | Stock full size ranges, move PPE stations to the point of use, and adjust the workflow so the safe way is the easy way | | The assessment document never changes any SOP | No owner assigned after sign-off | Map each control to a specific SOP line item with a named owner and a due date | ## Variations on the Standard Assessment **Task-based versus facility-based assessment.** A facility baseline covers the room, its systems, and its agents once. A task-based assessment re-scores individual procedures whenever they change. Mature programs run both, the baseline annually and the task layer continuously. **Quantitative modeling.** Two published approaches extend the standard process where consequence is high. CFD coupled with the Wells-Riley equation maps exposure risk across the actual room geometry and time course, turning disinfection placement and timing into calculated decisions rather than habits [1]. Separately, engineering methodology exists to calculate reverse airflow tolerance for BSL-3 and AII rooms, which quantifies how much airflow imbalance a room can absorb before containment is lost [2]. **Environmental risk management systems.** Institutions operating multiple high-level containment suites sometimes formalize the environmental half of the assessment into a management system with defined risk points, responsibilities, and emergency response processes, an approach described for BSL-3 and BSL-4 facilities [3]. **AI-assisted compliance monitoring.** Camera-based systems now add a verification layer for PPE and behavior. One published BSL-3 deployment trained a deep learning model on 4,112 images across 15 PPE compliance classes, covering clothing, masks, hoods, double-layer gloves, shoe covers, and respirators, and reached 97.52 percent accuracy and 97.03 percent recall in verifying entry compliance aligned with World Health Organization standards. An internal system monitoring protocol compliance in defined alert zones reached about 90 percent accuracy on 3,347 images [5]. Such systems supplement the human audit, they never replace it. **Program-level gap analysis.** Networks of BSL-3 facilities also assess the program rather than the room. The Research Alliance for Veterinary Science and Biodefense BSL-3 Network (RAV3N), working with ABSA International and sponsored by the USDA and the U.S. Department of State, ran a structured gap assessment workshop with experts from seven countries. Its findings grouped into applied biorisk gaps needing new experiments, completed research that was never published or shared, risk assessment tools and process enhancements still required, and literature summaries waiting to be written [6]. Borrowing that structure for an internal program review exposes blind spots a single-facility team cannot see. **Animal containment.** Work with infected animals runs under ABSL-3, animal biosafety level 3, where bedding handling, cage dumping, and animal movement add aerosol and bite pathways. The same step-by-step structure applies, with the procedure inventory expanded to every animal-care task. ## Records, Storage, and How Long an Assessment Stays Valid ![Decision path asking whether agents, procedures, engineering, staffing or incidents changed, leading to reassess now or wait for the annual review.](/article-images/illustrations/bsl-3-risk-assessment-out-of-cycle-triggers.webp) Store the signed assessment as a version-controlled document with a named owner, and keep superseded versions so the history of decisions survives staff turnover. Link the assessment to its dependent records: SOP revisions, training files, filter test reports, autoclave indicator logs, and incident reports. Paper copies kept in the laboratory office should be marked as controlled copies with a version number, because an outdated matrix in a binder is worse than no matrix at all. An assessment stays valid until something it assumed changes. Triggers for an out-of-cycle update: a new agent, a new aerosol-generating procedure, renovation or HVAC work, any filter replacement, a change in room staffing or skill mix, and any exposure or near miss. Absent a trigger, the annual review stands. Revalidate engineering controls after every filter change or airflow adjustment, because the published filter record shows that mounting and sealing faults can appear and be corrected between routine tests. ## Common Mistakes and Limitations The most common BSL-3 risk assessment mistakes follow a recognizable pattern, and every one of them is avoidable. 1. **Copying another facility's assessment.** Room geometry, HVAC design, and staff skill differ. A borrowed document inherits errors nobody in the building can explain. 2. **Scoring the room instead of the tasks.** A room-level score hides the single centrifugation step that carries most of the risk. Score tasks. 3. **Leaving support staff out.** Maintenance workers, custodial staff, and animal care staff enter the containment boundary. If the assessment does not name them, their training and PPE decisions have no anchor. 4. **Treating PPE as the primary control.** PPE sits last in the hierarchy for a reason. A program that reaches for respirators before it fixes airflow and work practices has the order inverted. 5. **Assuming filters perform without testing.** Filtration performs well when filters are installed and sealed correctly, and the published record includes a system in which some installed filters underperformed until faults at the cartridge and mounting bracket were found and corrected [4]. Test the installation, not the brochure. 6. **Running training once.** A signed attendance sheet from onboarding is not competency. Microbiology biosafety training needs direct observation and refreshers tied to SOP changes. 7. **Letting the document age.** An assessment older than the last renovation, the last new agent, or the last incident no longer describes the laboratory. 8. **Skipping the environmental and emergency walk.** Waste routes and failure scenarios feel peripheral until they become the main event. Managed frameworks exist precisely to keep this half of the assessment from being forgotten [3]. 9. **Treating matrix scores as measurements.** The score ranks hazards for discussion. It does not measure exposure, and two competent teams can defend different scores for the same hazard. 10. **Ignoring spatial variation inside the room.** Exposure risk is not uniform. Releases in recirculation zones near large equipment persist longer and expose staff more than releases near the exhaust, which is why the procedure inventory in step 3 records where each task happens. Beyond these mistakes, the method itself has limits. A biosafety risk assessment is structured judgment, not measurement, and its quality depends on the honesty of the walkthrough and the experience of the team. The field recognizes its own evidence gaps. A structured gap analysis across a network of veterinary and agricultural BSL-3 facilities concluded that risk assessment tools and process enhancements are still required, and that much completed biorisk research was never published or shared, leaving assessors to extrapolate [6]. Facility-specific decisions, including which agents a given suite may hold and which risks the institution accepts, belong to the institutional biosafety committee and the biosafety officer, who own the final document. ## Frequently Asked Questions ### What is a BSL-3 risk assessment? A BSL-3 risk assessment is the documented process of identifying how workers, the public, and the environment could be exposed to agents handled at biosafety level 3, rating each exposure pathway for likelihood and consequence, and selecting controls that reduce the remaining risk to a level the institution accepts in writing. ### How often should a biosafety risk assessment be repeated? At least annually, and out of cycle whenever agents, procedures, equipment, staffing, or the facility change, and after any exposure or near miss. Filter replacements and HVAC work also trigger revalidation of the affected engineering controls. ### Who should lead a BSL-3 risk assessment? The biosafety officer leads it, with the laboratory director, a daily-bench technologist, the facility or HVAC engineer, and occupational health at the table. Teams that omit the technologist or the engineer produce documents that describe a laboratory nobody actually works in. ### Can BSL-3 risk be quantified? Yes, for defined scenarios. The Wells-Riley model estimates infection probability from inhaled infectious dose, and computational fluid dynamics maps airflow through the actual room. Reserve these tools for the highest-consequence scenarios where their cost pays off. ### How is HEPA filtration performance verified? Through in-place leak testing of the installed filter together with its housing and seals, plus continuous pressure monitoring. Any decline from baseline triggers inspection of mounts and gaskets first, then cartridge replacement and retest. ### What should microbiology biosafety training cover? Agent hazards and exposure routes, the specific SOPs each person performs, spill and exposure response, and incident reporting, with competency confirmed by direct observation and refreshed when SOPs or agents change. ### What triggers an out-of-cycle reassessment? A new agent, a new aerosol-generating procedure, renovation or HVAC work, filter replacement, staffing changes that alter the skill mix, and any exposure incident or near miss. ### What are the most common BSL-3 risk assessment mistakes? Copying a template from another facility, scoring rooms instead of tasks, omitting support staff, treating PPE as the primary control, and allowing the document to age past the changes it is supposed to describe. ## Sources 1. [Potential infection risk assessment of improper bioaerosol experiment operation in one BSL-3 laboratory based on the improved Wells-Riley method](https://doi.org/10.1016/J.BUILDENV.2021.107974) 2. [Health and safety risk assessment methodology to calculate reverse airflow tolerance in a biosafety level 3 (BSL-3) or airborne infection isolation room (AII) environment](https://doi.org/10.1504/IJRAM.2010.035252) 3. [Environmental Risk and Assessment Management System of High-level Biosafety Laboratory.](https://pubmed.ncbi.nlm.nih.gov/32288926/) 4. [Assessment of the risk of infectious aerosols leaking to the environment from BSL-3 laboratory HEPA air filtration systems using model bacterial aerosols.](https://pubmed.ncbi.nlm.nih.gov/38620193/) 5. [Enhancing safety with an AI-empowered assessment and monitoring system for BSL-3 facilities.](https://pubmed.ncbi.nlm.nih.gov/39811271/) 6. [Findings and Recommendations of the RAV3N Applied Biorisk and Biosafety Gap Assessment Workshop.](https://pubmed.ncbi.nlm.nih.gov/40151386/)