# How to Set Up a Culture Collection Management System By the end of this guide you will have a working culture collection management system for a microbiology laboratory: an accession scheme that cannot collide, a data dictionary you can load into a spreadsheet or a small relational database, a physical storage map that tells you exactly which freezer, rack, box, and position holds each vial, a document set that survives an audit, and a verification and quality control loop that keeps identity, purity, and viability honest over years of storage. Have these items on hand before you start: a freezer inventory with rack and box counts, a spreadsheet application or the `sqlite3` command-line tool, cryovials rated for the storage temperature you plan to use, molecular-grade glycerol or dimethyl sulfoxide (DMSO), a label printer with cryogenic labels, access to an identification platform such as matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) or Sanger sequencing, and a second person who can act as reviewer. The whole build takes about one working week for a collection of a few hundred strains, plus a validation period. ## What a Culture Collection Management System Covers A culture collection is a curated set of microorganisms held for a defined purpose, with documented provenance, confirmed identity, and a known storage history. A management system is the combination of four layers that make that set usable: 1. **Physical layer.** Freezers, cryogenic tanks, lyophilized ampoules, cryovials, boxes, racks, and the labels on them. 2. **Data layer.** The accession register, the taxonomy table, the storage map, the verification log, and the transaction history. 3. **Document layer.** Standard operating procedures (SOPs), work instructions, forms, and the records those forms generate. 4. **Governance layer.** Named custodians, release authority, review cycles, change control, and audit. Vocabulary matters here, because ambiguous words cause most collection failures. An **accession** is a unique, permanent identifier assigned to a strain when it enters the collection. A **master stock** (also called a seed stock or primary stock) is the earliest, least-passaged preparation of that accession, held in reserve. A **working stock** is a daughter vial derived from the master stock and used for day-to-day work. **Passage number** counts how many times the organism has been subcultured since it was isolated or received, with the deposit preparation as passage zero. A **cryoprotectant** is a solute such as glycerol or DMSO that reduces ice damage during freezing. **Provenance** is the documented chain from original sample to the vial in your hand. The seed-stock concept is not an administrative nicety. In vaccine manufacturing, selecting and managing suitable seed strains according to standard principles is described as a critical factor in producing high-quality biological products, and complete genome data have been generated specifically so that manufacturers can confirm at the strain level whether the same seed strain is being used over time [1]. The same principle applies at laboratory scale: if you cannot demonstrate that the vial you pulled today is the same organism as the vial you deposited three years ago, you do not have a collection, you have a freezer full of unknowns. Genome-level records are now practical for small collections. Complete genomes have been produced for five bacterial strains recovered directly from live attenuated animal vaccines, including *[Erysipelothrix rhusiopathiae](/knowledge/bacteria/livestock-bacteria/erysipelothrix-rhusiopathiae)* NL-11, *[Clostridium chauvoei](/knowledge/bacteria/livestock-bacteria/clostridium-chauvoei)* B, two *[Bordetella bronchiseptica](/knowledge/bacteria/pet-bacteria/bordetella-bronchiseptica)* strains, and *Mycoplasmoides gallisepticum* str. F, with genome sizes of 1.78, 2.40, 5.34, 5.34, and 0.95 Mb using a hybrid of Illumina short reads and Oxford Nanopore long reads [1]. A parallel study produced complete, gap-free genomes for eleven strains used in inactivated animal vaccines, sizes 2.28 to 5.35 Mb, all with completeness above 99% and contamination below 1% [2]. Those numbers show what a reference-grade record looks like when a collection takes identity seriously. ## Governance: Scope, Roles, and Documents ### Decide what the collection is for Write a one-page scope statement before you touch a freezer. State the categories you will hold. A practical scheme uses five: - **REF**: reference strains used to validate methods and media. - **QC**: organisms with defined performance expectations for routine quality control. - **CLIN**: clinical or field isolates retained for investigation, teaching, or resistance surveillance. - **SEED**: production or vaccine seed strains held under master and working stock control. - **ENV**: environmental and equipment monitoring isolates. The category drives everything downstream: storage temperature, passage limits, release authority, and destruction rules. A QC organism that fails its expected reaction is a finding. A CLIN isolate that fails identification is a data point. ### Assign ownership and custodianship Name a **collection curator** who owns the register and approves deposits and withdrawals. Name a **custodian** for each storage unit who holds the physical key or access credential. Name a **reviewer** who is not the curator and who signs off on identity verification and release. Three names, three signatures, no shared logins. ### The minimum document set A collection that passes review has these documents on day one: - Collection scope and policy (one to three pages). - Accession numbering SOP. - Deposit and withdrawal SOP. - Cryopreservation and recovery SOP, one per storage format. - Identity verification SOP, one per method. - Storage map maintenance SOP. - Freezer alarm response and disaster recovery SOP. - Destruction and transfer SOP, including documentation for regulated material. - Training record and competency matrix. - Change control form. Each SOP carries a document number, a version, an effective date, an author, and an approver. The format matters less than the presence of those six fields on every page. ## The Setup Walkthrough ### Step 1: Build the folder skeleton and audit what you already have Create a single root folder with a fixed structure. Keep it on a backed-up network share, not on a laptop. ```bash mkdir -p collection/{sop,records,storage_maps,verification,backups} ls -1 collection ``` Expected output: ``` backups records sop storage_maps verification ``` Now audit. Pull every box from every freezer onto a bench sheet with four columns: freezer, rack, box, position. Record what the label says, not what you think it says. Expect a 10 to 30 percent discrepancy rate in a collection that has never been audited. Vials with illegible labels go into a quarantine box and are re-identified before anything else happens. ### Step 2: Design an accession number that cannot collide Use a structured identifier with a fixed field order. A workable format is `PREFIX-YEAR-CATEGORY-SEQUENCE-CHECK`, for example `ZK2026QC00001-1`. The prefix identifies the site or collection, the year identifies the intake year, the category matches your five-letter scheme, the sequence is a zero-padded counter that never resets, and the check digit catches transcription errors at the bench. ```python import re def accession(prefix, year, category, seq): body = f"{prefix}{year}{category}{seq:05d}" digits = [int(c) for c in re.sub(r"\D", "", body)] check = sum(d * (i + 1) for i, d in enumerate(digits)) % 10 return f"{body}-{check}" print(accession("ZK", 2026, "QC", 1)) print(accession("ZK", 2026, "QC", 2)) ``` Expected output: ``` ZK2026QC00001-1 ZK2026QC00002-2 ``` Two rules make this scheme safe. First, sequence numbers never reset, not even at year boundaries, so a retired number is never reused. Second, the check digit is recalculated at label printing, so a transposed digit is caught before the vial enters the freezer. ### Step 3: Write the data dictionary before you touch a freezer The data dictionary is the list of fields, their allowed values, and their required status. Write it as a table first, then convert it to a schema. This is the step most collections skip, and it is the reason their spreadsheets collapse at around 500 accessions. ```sql CREATE TABLE strain ( accession TEXT PRIMARY KEY, taxon TEXT NOT NULL, strain_label TEXT, source TEXT NOT NULL, isolation_date TEXT, received_date TEXT, category TEXT NOT NULL CHECK (category IN ('REF','QC','CLIN','SEED','ENV')), passage_at_deposit INTEGER NOT NULL DEFAULT 0, storage_id TEXT NOT NULL, custodian TEXT NOT NULL, status TEXT NOT NULL DEFAULT 'QUARANTINE' CHECK (status IN ('QUARANTINE','RELEASED','RESTRICTED','DESTROYED')) ); CREATE TABLE storage ( storage_id TEXT PRIMARY KEY, freezer TEXT NOT NULL, rack INTEGER NOT NULL, box INTEGER NOT NULL, position TEXT NOT NULL, UNIQUE (freezer, rack, box, position) ); CREATE TABLE verification ( verification_id INTEGER PRIMARY KEY, accession TEXT NOT NULL REFERENCES strain(accession), method TEXT NOT NULL CHECK (method IN ('MALDI','SANGER16S','WGS','PHENO')), performed TEXT NOT NULL, result TEXT NOT NULL, concordant INTEGER NOT NULL CHECK (concordant IN (0,1)), operator TEXT NOT NULL ); ``` Every free-text field is a future inconsistency. Constrain `taxon` with a controlled vocabulary derived from a current nomenclature source, constrain `source` with a list of allowed provenance types, and store dates as ISO 8601 strings (`YYYY-MM-DD`) so sorting works. Common errors at this stage and their exact messages: - `Error: near "CREAT": syntax error` means a truncated keyword. Retype the statement. - `Parse error near line 12: no such table: storage` means the parent table was never created. Run the `CREATE TABLE` statements in dependency order. - `Runtime error near line 3: CHECK constraint failed: category IN ('REF','QC','CLIN','SEED','ENV')` means a value outside the list was inserted. Fix the input, not the constraint. ### Step 4: Create the storage map The storage map is a coordinate system, not a description. Use a fixed nesting order: freezer, rack, box, position. A single vial therefore has the address `F1-R03-B12-C4`, which reads as freezer 1, rack 3, box 12, position C4. Print that string on the label and store it verbatim in the `storage` table. Do the capacity arithmetic once. A 9 by 9 cryobox holds 81 vials. Twenty boxes per rack gives 1,620 vials per rack. Four racks per freezer gives 6,480 vials per freezer. If your policy is three vials per accession (one master, two working), that freezer holds 2,160 accessions before you need to expand. ```sql INSERT INTO storage VALUES ('ZK2026QC00001-1','F1',3,12,'C4'); INSERT INTO storage VALUES ('ZK2026QC00002-2','F1',3,12,'C5'); INSERT INTO storage VALUES ('ZK2026QC00003-3','F2',1,3,'A1'); ``` If you reuse a coordinate, sqlite returns: ``` Runtime error near line 1: UNIQUE constraint failed: storage.freezer, storage.rack, storage.box, storage.position ``` The fix is to free the position in the register first, then reinsert. Never overwrite a coordinate silently. ### Step 5: Prepare cryopreservation and recovery protocols Two formats cover most laboratory collections: glycerol stocks at -80 °C and lyophilized ampoules at room temperature or 4 °C. A third, vapor-phase liquid nitrogen, is used for organisms that lose viability at -80 °C. For glycerol stocks, the final concentration in the vial, not the concentration in the bottle, is what matters. A common working target is 15 percent glycerol by volume. If you prepare a 30 percent glycerol solution and mix it one part to one part with a fresh broth culture, the final concentration is 15 percent. ```text 30% glycerol stock : culture = 1 : 1 -> 15% final glycerol ``` For DMSO, a final concentration of 5 to 10 percent is typical. Add the cryoprotectant to the cell suspension, not the reverse, and dispense within 30 minutes. Freeze at approximately -1 °C per minute using a controlled-rate device or an insulated box, then transfer to the final storage temperature after at least two hours. The recovery protocol is the other half of the SOP. Thaw rapidly in a 35 to 37 °C water bath, wipe the vial with 70 percent ethanol, and subculture immediately. Record the recovery date and the new passage number. An organism recovered from a master stock at passage 0 becomes passage 1. ### Step 6: Fix the identity verification workflow Identity verification answers two different questions with two different methods. Genus and species confirmation is routine. Strain-level confirmation is not. For routine identification, MALDI-TOF MS is the workhorse. In a comparison of 292 isolates from an academic laboratory and outside laboratories, MALDI-TOF MS disagreed with commercial phenotypic systems for one bacterial and three yeast isolates, and the discrepant results matched the API system with 100 percent correlation after resolution [3]. That level of agreement supports routine use. The same study found that from four polymicrobial blood cultures, MALDI-TOF MS identified only one organism, and that 21 *Escherichia coli* isolates were reported as *E. coli/Shigella* [3]. Both findings have direct collection implications: purity must be confirmed before MALDI-TOF is trusted, and the *E. coli/Shigella* report is a genus-level answer that cannot be used to close a species-level record. Instrument choice affects throughput and cost but not the underlying rule. In a side-by-side evaluation of 154 bacterial and yeast isolates, the VITEK MS PRIME identified 96 percent with the PICKME nib and 95 percent with a loop to genus level, compared with 99 percent for the Bruker Biotyper [4]. On short-incubation positive blood culture isolates, species-level identification was 84 percent for Biotyper, 80 percent for PRIME with PICKME, and 81 percent for PRIME with a loop [4]. For a collection, the practical takeaway is that deposit-grade isolates should be grown under routine conditions before verification, not from short incubation. For strain-level confirmation, [whole genome sequencing](/blog/guides/whole-genome-sequencing) is the reference method. The eleven-strain vaccine dataset with completeness above 99 percent demonstrates what is achievable [2], and the five-strain live vaccine dataset shows the same workflow applied to organisms with very different genome sizes [1]. Where sequencing is unavailable, multiple-locus variable-number tandem repeat analysis (MLVA) can detect small genomic changes. MLVA with 16 loci applied to 63 commercial batches of *Brucella abortus* S19 and RB51, plus ten in vitro serial passages of reference strains, showed that both vaccine strains were genetically stable and homogeneous within their groups, yet batches from one manufacturer of each strain carried a one-repeat difference at locus Bruce07 relative to the reference [5]. That is the correct role for a typing method in a collection: not routine identification, but periodic confirmation that a seed stock has not drifted. ### Step 7: Set release criteria and a stability monitoring schedule A vial moves from `QUARANTINE` to `RELEASED` only when all four criteria are met and documented: 1. **Purity.** A single colony morphology on primary plating, Gram stain consistent with the taxon, and no mixed MALDI-TOF spectrum. 2. **Identity.** Species-level agreement with the deposit record by an approved method, with the method named in the verification table. 3. **Viability.** Growth within the expected time after recovery from storage. 4. **Provenance.** Source, date, and depositor recorded and traceable. Release is a signature, not a status change in a database. The reviewer signs the verification record, and the curator changes the status. Two people, two steps. Set a stability monitoring schedule at the same time. A workable default is to recover and re-verify the oldest vial of each accession at 12 months, then at 24 months, then every 24 months thereafter. Any accession with a stability failure triggers a review of the whole storage position, because a failing vial is often a failing box. ### Step 8: Build or configure the database For fewer than 500 accessions and one or two users, a versioned spreadsheet is adequate if it has one sheet per table, no merged cells, no color coding as data, and a locked header row. Above that threshold, use a relational database. The schema in Step 3 loads directly. ```bash sqlite3 collection/records/collection.db < collection/records/schema.sql sqlite3 collection/records/collection.db ".tables" ``` Expected output: ``` storage strain verification ``` Now populate and query a release list: ```sql SELECT s.accession, s.taxon, st.freezer || '-R' || printf('%02d', st.rack) || '-B' || printf('%02d', st.box) || '-' || st.position AS location FROM strain s JOIN storage st ON s.storage_id = st.storage_id WHERE s.status = 'RELEASED' ORDER BY s.accession LIMIT 3; ``` Expected output: ``` ZK2026QC00001-1|Escherichia coli|F1-R03-B12-C4 ZK2026QC00002-2|Staphylococcus aureus|F1-R03-B12-C5 ZK2026CL00001-1|Pseudomonas aeruginosa|F2-R01-B03-A1 ``` The most common query error in new collections is `no such column: st.storage_id`, which means the join key was spelled differently in the two tables. Pick one spelling and enforce it in the data dictionary. ### Step 9: Back up, version, and test restore Run a backup on a schedule and a restore test on the same schedule. An untested backup is a hope, not a control. ```bash find collection -type f -name "*.db" -print0 | xargs -0 sha256sum > collection/backups/manifest.sha256 sha256sum -c collection/backups/manifest.sha256 ``` Expected output: ``` collection/records/collection.db: OK ``` A mismatch prints: ``` collection/records/collection.db: FAILED sha256sum: WARNING: 1 computed checksum did NOT match ``` A failure means the file changed after the manifest was written, which is expected during working hours and unexpected after a restore. Rebuild the manifest immediately after any approved restore and record the restore in the change log. ### Step 10: Validate, train, and schedule the audit Validation for a collection system is a set of challenge tests, not a document. Insert a deliberate duplicate coordinate and confirm the database rejects it. Print a label from a record and confirm the printed string matches the stored string. Pull three random vials and confirm the physical label matches the register. Recover one vial and confirm the identity result matches the deposit record. Training is competency-based. Each operator demonstrates accessioning, labeling, cryopreservation, recovery, and query entry under observation, and the observation is signed and dated. The audit cycle runs annually and covers a random sample of at least 10 percent of accessions plus every storage unit. ## Key Options at a Glance | Element | Lightweight option | Scaled option | Choose when | |--|--|--|--| | Data platform | Versioned spreadsheet, one sheet per table | Relational database or [laboratory information management system](/knowledge/bioinformatics/lims-explained-lab-information-management-systems) | More than 500 accessions or more than two concurrent users | | Storage format | Glycerol stock at -80 °C | Lyophilization or vapor-phase liquid nitrogen | Frequent access versus long-term archival | | Identity method | MALDI-TOF MS | 16S rRNA [gene sequencing](/knowledge/molecular-biology/gene-sequencing) or whole genome sequencing | Routine species confirmation versus novel taxa or strain-level claims | | Accession format | `PREFIX-YEAR-CAT-SEQ` | Add a check digit and barcode printing | Single site versus multi-site or automated labeling | | Passage control | Record passage at deposit | Cap passage number and re-derive from master stock | Research use versus production or reference use | | Backup | Weekly copy to a second device | Nightly copy plus off-site copy plus restore test | Small collection versus audit-facing collection | ## A Reusable Record Template The record format below works as a paper form and as a spreadsheet layout. One row per vial, one row per transaction. Copy the column names exactly so the data loads into the schema without renaming. | Column | Type | Allowed values or format | Required | |--|--|--|--| | accession | text | `PREFIX-YEAR-CAT-SEQ-CHECK` | yes | | taxon | text | controlled vocabulary, genus and species | yes | | strain_label | text | free text, no spaces | no | | source | text | controlled list (clinical, environmental, vaccine, reference, other) | yes | | isolation_date | date | `YYYY-MM-DD` | yes | | received_date | date | `YYYY-MM-DD` | yes | | category | text | REF, QC, CLIN, SEED, ENV | yes | | passage_at_deposit | integer | 0 or greater | yes | | storage_id | text | matches storage table | yes | | freezer, rack, box, position | text or integer | fixed order | yes | | verification_method | text | MALDI, SANGER16S, WGS, PHENO | yes | | verification_date | date | `YYYY-MM-DD` | yes | | verification_result | text | reported organism | yes | | concordant | boolean | 0 or 1 | yes | | custodian | text | named person | yes | | status | text | QUARANTINE, RELEASED, RESTRICTED, DESTROYED | yes | | transaction_type | text | deposit, withdraw, transfer, destroy, verify | yes | | transaction_date | date | `YYYY-MM-DD` | yes | | operator | text | named person | yes | A worked example row for a QC organism: ```text ZK2026QC00001-1 | Escherichia coli | K12-derivative | reference | 2026-01-14 | 2026-01-16 | QC | 0 | F1-R03-B12-C4 | MALDI | 2026-01-20 | Escherichia coli | 1 | A. Patel | RELEASED | deposit | 2026-01-16 | A. Patel ``` This is the "format" question answered concretely: fixed field order, controlled vocabularies, dates in ISO 8601, one row per transaction, and no free text in any field that a query will filter on. ## Microbiology Quality Control Tied to the Collection The collection exists to serve microbiology quality control, so the two systems must share records. Quality control in a microbiology laboratory rests on organisms with predictable, documented behavior. A QC organism without a verified identity, a known passage number, and a documented storage history cannot support a control decision. Three controls connect the collection to laboratory quality management: **Reference organism control.** Each QC accession carries an expected reaction profile for the tests it supports. When a new lot of medium, a new antimicrobial panel, or a new identification instrument is introduced, the reference organism is recovered from the collection and tested. The result is recorded against the accession, not against the lot alone, so the history of the organism is visible in the same place as the history of the reagent. **Passage control.** Organisms can lose plasmids, capsule, or specific antigens after repeated subculture. Cap the passage number for QC and production strains, and re-derive the working stock from the master rather than continuing to passage. The *Brucella* dataset shows why periodic confirmation matters: even genetically stable vaccine strains showed single-repeat differences at one locus in batches from particular manufacturers after routine production [5]. A collection that never re-checks will never detect that kind of drift. **Manufacturing quality control.** Where a collection supplies strains for biological product manufacture, the quality control expectations are explicit. A position statement on probiotic use in clinical practice developed twelve statements and recommendations covering quality control in the manufacturing of probiotics, criteria for selecting probiotics, and evidence for their use in diarrheal disease, and it framed efficacy as strain-specific [6]. A collection that cannot demonstrate strain identity cannot support a strain-specific claim. Whole genome sequencing is the strongest available identity control for seed and reference material. Complete genomes with high completeness and minimal contamination have been generated for strains used in inactivated vaccines, with the explicit purpose of confirming at strain level whether manufacturers are using the same seed strains over time [2]. For live attenuated vaccine strains, the same approach provides a reference against which production seed stocks can be compared [1]. A small laboratory can adopt the principle without the full workflow: store the raw reads or the assembled genome alongside the accession record, and use it as the reference for every future identity check of that accession. Instrument selection for routine QC follows the same logic. MALDI-TOF MS gives fast species-level answers with high concordance to conventional systems [3], and current platforms differ in throughput and workflow rather than in the underlying principle, with genus-level agreement between 95 and 99 percent across the systems compared in one evaluation [4]. The collection record should name the method and the instrument, because a result produced on one platform is not automatically interchangeable with a result produced on another. ## Common Mistakes and Limitations **Storing location and identity in one cell.** `E. coli freezer 1 box 3` cannot be queried, sorted, or joined. Split every attribute into its own column. **Free-text taxonomy.** `E. coli`, `Escherichia coli`, `E.coli`, and `eco` are four different strings. Use a controlled vocabulary and a pick list. **No passage field.** Without a passage number, you cannot tell a fresh recovery from a vial that has been subcultured twelve times. **Master and working stocks in the same box.** A single freezer failure destroys both. Separate them physically and, where possible, put the master stock in a different storage unit. **Releasing before quarantine ends.** A vial that has not completed purity, identity, viability, and provenance checks is not a collection strain. It is a candidate. **Labeling with a permanent marker.** Marker ink fails at -80 °C and in liquid nitrogen vapor. Use cryogenic labels printed from the register, never handwritten from memory. **Trusting MALDI-TOF on a mixed culture.** MALDI-TOF MS identified only one organism from four polymicrobial blood cultures in one evaluation [3]. Confirm purity on a plate before you put a target on the instrument. **Treating a genus-level report as a species-level answer.** The *E. coli/Shigella* report is a real and expected limitation of the method [3]. Record it as genus-level and escalate if the record requires species. **Assuming a commercial library covers your organism.** Libraries are curated for clinically common organisms. Unusual environmental or veterinary isolates may score low or not at all. **No alarm or no response plan.** A freezer alarm without a named responder and a documented action list is decoration. **Duplicate accessions after a re-import.** Import into a staging table, run a duplicate query on the accession and on the coordinate pair, then merge. **No change control.** A schema change without a version number and a migration script will silently break queries months later. The limitations are real. This system is designed for a single laboratory or a small network, not for a national repository with legal deposit obligations. Identification methods have finite resolution, and a species-level call from MALDI-TOF MS is not a strain-level call. Genome-based confirmation requires bioinformatics capacity that many service laboratories do not have in-house. For clinical isolates connected to animal cases, the attending veterinarian and the receiving diagnostic laboratory should agree on retention and release before any strain is banked. ## Frequently Asked Questions ### How many vials per strain should a new collection bank? Bank at least three: one master stock that is never opened for routine work, and two working stocks. For production or reference strains, bank five to seven and store the master in a separate storage unit. ### What is the difference between a master stock and a working stock? A master stock is the earliest, least-passaged preparation of an accession, held in reserve. A working stock is a daughter vial derived from it for daily use, and it carries a higher passage number. ### Can a spreadsheet serve as a culture collection management system? Yes, for fewer than 500 accessions and one or two users, provided it has one sheet per table, no merged cells, locked headers, and a documented backup schedule. Above that size, move to a relational database. ### What temperature should I store cultures at? Glycerol stocks are conventionally held at -80 °C, lyophilized ampoules at 4 °C or room temperature, and organisms that lose viability at -80 °C in vapor-phase liquid nitrogen. The choice is organism-specific and should be recorded per accession. ### How often should I re-verify stored strains? A common default is 12 months for the oldest vial of each accession, then 24 months thereafter. Any viability or identity failure triggers a review of the surrounding storage position. ### What is a passage number and why does it matter? A passage number counts subcultures since isolation or receipt, with the deposit preparation as passage zero. It matters because repeated passage can change phenotype, plasmid content, and antigen expression. ### Do I need whole genome sequencing for a small collection? No. MALDI-TOF MS or 16S rRNA sequencing supports routine species-level verification. Sequencing becomes valuable for seed stocks, production strains, and any accession where a strain-level identity claim must be defended. ### What should I do if MALDI-TOF gives a discordant identification? Re-purify the culture, repeat the target preparation, and confirm with a second method such as 16S rRNA gene sequencing. Record both results in the verification table with the concordance flag set to 0, and do not release the vial until the discrepancy is resolved. ## Sources 1. [High-quality genome assembly and annotation of live animal vaccine bacteria strains in South Korea.](https://pubmed.ncbi.nlm.nih.gov/40898044/) 2. [High-quality genome assembly and annotation of inactivated animal vaccine bacteria strains in South Korea.](https://pubmed.ncbi.nlm.nih.gov/42576204/) 3. [Use of matrix-assisted laser desorption/ionisation-time of flight mass spectrometry analyser in a diagnostic microbiology laboratory in a developing country.](https://pubmed.ncbi.nlm.nih.gov/29435422/) 4. [Performance and workflow comparison of the VITEK MS PRIME and Bruker Biotyper MALDI-TOF MS systems.](https://pubmed.ncbi.nlm.nih.gov/40539832/) 5. [Genetic stability of Brucella abortus S19 and RB51 vaccine strains by multiple locus variable number tandem repeat analysis (MLVA16).](https://pubmed.ncbi.nlm.nih.gov/23933375/) 6. [Use of probiotics in clinical practice with special reference to diarrheal diseases: A position statement of the Malaysian Society of Gastroenterology and Hepatology.](https://pubmed.ncbi.nlm.nih.gov/33490608/)