Glossary

Plain-language definitions of the terms you meet across PlasticDB's pages, downloads and tools.

Draft wording — pending maintainer review. These definitions were compiled from PlasticDB's existing documentation pages and design notes, and have not yet been checked line by line by the maintainer.

How evidence is recorded

Evidence spine
The chain of records that every degradation claim hangs from: paper → interaction → experiment → observation. PlasticDB never stores “organism X degrades plastic Y” as a bare fact; each claim is anchored to the evidence behind it. See Data Model.
Paper (reference)
The published source a record comes from, usually identified by its DOI. Papers are never linked directly to an organism or protein: a paper's organisms and proteins are worked out from the interactions it supports. See Reference (paper).
Interaction
The basic unit of evidence: one actor tested against one plastic, as reported by one paper, together with that paper's overall result. An organism tested against three plastics is three interactions. See Interaction.
Actor
The thing that was tested against the plastic. Every interaction has exactly one actor, which is a microorganism, a protein or a protein variant. A mixed community (a consortium, sludge or enrichment culture) is never an actor. See Actors.
Experiment
One method or assay run under stated conditions (temperature, pH, duration and so on), attached to an interaction. A paper that probes the same pairing with both SEM and weight loss records two experiments. See Experiment.
Observation
One measurement an experiment produced, such as a weight-loss percentage or a description of surface pitting. One experiment can yield several observations, for example weight loss at 2, 4 and 6 weeks. See Observation.
Method
The technique used in an experiment (for example weight_loss, sem or hplc), chosen from a fixed list. It says how something was measured. See Experiment methods.
Measurement type
What an observation quantified (for example weight_loss, co2_evolved or monomer_released), as distinct from the method that produced it. See Measurement types.
Controlled vocabulary
A fixed list of allowed values for a field, such as methods, measurement types, results and environments. Using one list everywhere keeps records comparable across studies. See Controlled Vocabularies.
Enzyme kinetics and properties
Optional biochemical values an enzyme experiment may report: kinetic constants (Km, kcat, Vmax), and properties such as optimum temperature and pH, melting temperature (Tm) and thermal half-life. They are recorded under the experiment they were measured in, because the same enzyme can have different values under different conditions. See Enzyme kinetics & properties.

Plastics and compounds

Plastic
The polymer an actor was tested against, held as one canonical record per plastic (PET, PHB, PBAT and so on) and named by its standard abbreviation where one exists.
Homopolymer, copolymer, blend and formulation
The kinds of plastic record. A blend (for example a commercial PBAT-based grade) is its own plastic record, so a result on a blend is never counted towards any single constituent: “degraders of PLA” counts pure PLA only. A formulation is one polymer plus non-polymer additives, such as plasticised PVC.
Substrate composition
For a blend or copolymer, the constituents and their fractions as actually tested in one experiment (for example a PLA/PCL blend at 60:40). Ratios vary between studies, so they are recorded per experiment. See Substrate composition.
Substrate form
The physical form of the plastic used in an experiment, such as amorphous film, crystalline film, powder or whole polymer. See Other controlled fields.
Compound
A chemical that is not itself a plastic: monomers, additives such as plasticisers, and degradation or pathway intermediates. An actor that acted only on a compound derived from a plastic (a monomer, MHET or BHET, for example) records no interaction with that plastic. See rule R-16.

Organisms and strains

Microorganism
A microbe (bacterium, fungus or archaeon) recorded at species level, so that ten strains of one species count once. An organism the paper identified only to genus (“Rhodococcus sp.”) is stored at the genus, and the paper's own name is kept as a synonym so search still finds it.
Strain (strain designation)
The isolate or strain label a paper uses, such as “201-F6”. Because the same species recurs across papers with different isolates, the strain is recorded with that paper's evidence rather than on the species record.
Strain occurrence
A microorganism strain or isolate as reported in one study: one record per paper × species × strain. It is the organism actor behind an interaction and holds the study-specific details: strain designation, genome assembly, culture-collection numbers and where it was isolated.
Sample source (isolation environment)
Where a microorganism was isolated in a particular study: the broad environment type (soil, compost, marine water, landfill and so on), country, region, coordinates and a description of the sample. A species can carry several sources, each tagged with its paper. See Sample source.
Tax ID (NCBI taxonomy)
The numeric identifier of an organism in the NCBI Taxonomy database, shown as “Tax ID” on microorganism pages and linking to NCBI. It anchors the organism's lineage; an ambiguous organism name is refused rather than guessed.
Lineage
The organism's chain of taxonomic ranks (domain, phylum, class, order, family, genus), fetched from NCBI using the tax ID rather than entered by hand, and refreshed weekly. It drives the lineage breadcrumb on microorganism pages and the taxonomy tree.
Growth temperature class
A grouping by the organism's recorded optimum growth temperature: psychrophile (below 20 °C), mesophile (20 to below 45 °C) or thermophile (45 °C or above). It is not the assay temperature, and organisms with no recorded optimum fall into none of the three.
Strain type and constructs
The kind of strain a paper tested: wild type, expression host, knockout or deletion mutant, engineered strain, or not classified (the default). Results from expression hosts, knockouts and engineered strains stay visible and labelled on paper and protein pages but are never counted towards the species. See Knockouts, expression hosts and engineered strains.

Proteins and variants

Protein (enzyme)
A characterised enzyme, identified by a name plus an accession. A cell-free assay of a purified enzyme is a claim about the protein, recorded separately from any claim about the organism it came from. Protein names are not unique (many records are called “PETase”), so always identify a protein by its accession.
Wild-type protein
A protein as it occurs naturally, as opposed to an engineered variant. Protein lists and exports can show wild-type proteins, engineered variants, or both. See Wild-type and engineered.
Protein variant (engineered variant)
An engineered or otherwise modified version of a parent protein, such as a stabilised PETase mutant. It is a record in its own right, with its own accession, mutations, kinetics and papers, and is never folded into its parent. Variants are browsed through the Engineering view.
Accession
A stable identifier for a record. PlasticDB gives each protein and variant its own accession, “PB” plus four characters, assigned at random and never reused. Proteins also carry external accessions from GenBank/RefSeq, UniProt or the PDB (structures), which PlasticDB uses to fetch the sequence and other annotations.
EC number
The Enzyme Commission number classifying the reaction an enzyme catalyses (for example EC 3.1.1.101). PlasticDB derives it from the protein's identifiers rather than having contributors type it in.
Signal peptide
A leading segment of a protein's deposited sequence that is cut off to give the mature protein; the point where it is cut is the cleavage site. PlasticDB derives it for stored proteins, and the gene and genome annotation tools can predict it for your sequences with DeepSig. See Signal peptides.
Precursor and mature sequence
The stored sequence is the deposited, full-length precursor exactly as published; the mature sequence is that chain with the signal peptide removed, and PlasticDB derives it. Papers usually number mutations on the mature protein, which is why a variant can carry a numbering offset. See Protein variant actor.

Results and counts

Result (positive, negative, inconclusive)
The reported outcome of testing an actor against a plastic. Positive: degradation was observed. Negative: the actor was tested and no degradation activity was detected. Inconclusive: the test was run but the outcome was ambiguous or could not be clearly interpreted. Computational predictions are not recorded at all. See Degradation results.
Degrading
On a plastic page, the table of actors with at least one positive result for that plastic. An actor is placed in the best table it qualifies for, so one positive is enough even if other papers disagree. See “Degrading”, “inconclusive”, “no activity”.
Tested — no activity
The table for actors that were tested against a plastic one or more times where every result was negative. Negative results are kept because knowing what did not work avoids repeated experiments.
Mixed
A badge on microorganism and protein pages marking a pairing where papers report more than one outcome. The small tallies beside a result count the supporting papers on each side (for example 3+, 1?, 2−). See When papers disagree.
Degrader counts
Headline numbers across the site count degraders only: distinct microorganisms or proteins with at least one positive result. Actors whose only evidence is negative or inconclusive are left out of these totals but remain listed and searchable. See What the counts mean.
Whole organism vs. purified enzyme
A purified enzyme degrading a plastic does not, on its own, show that the organism it came from does. The two are recorded as separate interactions, each standing on its own evidence. See Actors.
Evidence scope (“via their enzyme”)
How organism counts and lists credit a microorganism with a plastic. By default (“Organism or its enzyme”) an organism is credited if it was assayed whole or if an enzyme purified from it was. “Tested whole only” and “Only via its enzyme” split that default into two halves that add up to it; the landing page's “via their enzyme” figure is the second half, organisms never themselves tested for that plastic.

Curation and releases

Extraction
A structured reading of a published paper produced by PlasticDB's separate AI extraction pipeline: the interactions, experiments and observations the paper reports, in the same shape a person uses on the Submit Data form. An extraction is reviewed by a curator before anything from it is written to the database.
Ingestion (paper ingestion)
One run of bringing a paper into PlasticDB, whether from an AI extraction or from a person's submission. It moves through download, extraction and review, and ends approved or rejected; a paper can have several runs over time.
Extraction confidence
The AI pipeline's own grade (high, medium or low) of how complete and unambiguous its extraction was. It grades the extraction, not the paper, and a record typed by a person has none. See Extraction confidence.
Submission
Records a registered user contributes from a paper through the Submit Data form or the API. Each submission is reviewed before it enters the public database, and accepted records are published under the site's licence with credit to the contributor. See Contributing Data.
Curator
A reviewer who checks extractions and submissions against the paper and approves, corrects or rejects them. Curators may approve AI extractions of published papers; accepting another person's submission is reserved for a superuser. Curators never publish records directly: approved records wait for the weekly release. See The review workflow.
Revision
A corrected version of an already approved paper: the curator reopens it and approves a revision, and the earlier run is marked superseded. This is distinct from rejecting a pending extraction and from withdrawing an approved paper, which deletes its records.
Staged
The state of a curated record that has been approved but not yet released. Staged records do not appear on public pages, searches, downloads, the sequence-search databases or release snapshots until the next weekly release publishes them.
Weekly dataset release
The scheduled weekly step (Thursdays, 06:20 UTC) that publishes everything staged, rebuilds the reference databases the sequence tools search, and cuts a snapshot when the data changed. All curated data becomes public this way; there are no embargoes or private deposits. See Which version of the data did you use?
Data version
The name of a dataset release: the ISO year and week, such as 2026.39, with a suffix such as 2026.39.2 for an extra release in the same week. It says which data you searched, and is deliberately different from the application version, which says which code is running. Quote it in your methods section.
Snapshot
An archived copy of the released dataset. A release cuts one only when the published data actually changed, so a quiet week gets none. Where a snapshot has been archived with a DOI, the releases endpoint gives it with a ready-made citation; otherwise cite the data version.
Merged record
A duplicate that has been folded into another record (the survivor). Every piece of evidence moves to the survivor, the old name is kept as a synonym, and the old accession redirects to the survivor so existing citations still work. See Duplicates and renames.
Tombstone
A retired record that is kept rather than deleted, and hidden from every public page, download and release. Its accession is never reused; a tombstoned accession reports that the record is gone.

Sequence-search tools

DIAMOND
A fast alternative to BLAST that most of PlasticDB's tools use to align your sequences against the PlasticDB enzyme database. A query reports at most its 25 best database matches. See Sequence-based tools.
BLASTP and BLASTX
The two search modes: BLASTP for protein input, BLASTX for nucleotide input, which is translated in all six reading frames. Your file must match the mode you choose; a mismatch is refused at submission.
Hit
A match between one of your sequences and a PlasticDB record. A hit is a similarity match, not a measurement: it suggests a candidate to test, not proof that your sequence degrades anything.
E-value
The significance threshold for reporting a hit; lower is more stringent. The default is 1e-5 on every tool page.
Percent identity (pident)
The percentage of identical amino acids in the aligned region between your sequence and the PlasticDB match. The searches use a minimum of 30% by default; a low-identity hit (30–40%) may not indicate genuine plastic-degrading activity.
Query coverage (qcov) and subject coverage (scov)
How much of each sequence the alignment spans: query coverage is the share of your sequence, subject coverage the share of the PlasticDB reference. Results pages first hide hits covering less than 70% of the reference; Show all brings them back without a rerun. See Default view.
Bitscore
The alignment score DIAMOND reports for each hit, listed in the results table and downloads. A minimum bitscore can be set to filter the results view; it is off by default.
Profile HMM (hidden Markov model)
A statistical summary of an aligned family of related sequences, rather than of a single sequence, so it captures what the family has in common. PlasticDB builds its profiles from families of curated plastic-degrading enzymes; a profile names no record and reproduces no sequence. See HMM Screen.
HMMER and the HMM Screen
HMMER is the software (hmmsearch) that searches protein sequences against profile HMMs. The HMM Screen tool uses it to find sequences resembling a degrading-enzyme family even when they have no close match in the database, for example predicted proteins from a metagenome.
GA / TC / NC cut-offs
Full-sequence bit-score cut-offs calibrated for each profile and stored in it, so anyone running hmmsearch on the library inherits them. TC (trusted cut-off) is the score of the lowest-scoring true member of the family. NC (noise cut-off) is the highest score any hard negative reached, a look-alike enzyme from a confusable family; a profile no hard negative scored against has no NC. GA (gathering cut-off) is set between the two by cross-validation. The HMM Screen applies GA by default (--cut_ga) and shows it beside each hit's score; its exploratory mode filters by E-value only. See Thresholds.

Pathways and trees

Pathway
A degradation pathway for a plastic: the ordered series of enzymatic steps that break it down. The Pathway Analysis tool currently maps your sequences onto the Pseudideonella sakaiensis PET-degradation pathway, ending at protocatechuate ring cleavage.
Reaction step
One enzymatic conversion within a pathway, with its EC number. In Pathway Analysis a step is called present only when a hit clears all of the identity, E-value and coverage thresholds (by default ≥40% identity, E-value ≤1e-10, ≥70% of the reference and ≥50% of your protein); missing steps are named with their best failing hit.
Taxonomy tree (cladogram)
A tree of PlasticDB microorganisms grouped by the ranks of their stored NCBI lineage. It is not a phylogenetic tree: no sequences are compared and branch lengths mean nothing. See Taxonomy Tree.