AptaGraph

About

What is in here, how to read it, and where it stops

What is in here

7,821 aptamers from 8,986 records in 5 public aptamer databases, linked to what they bind, to the measurement and conditions each record gives, and to the source row each statement comes from.

Source databaseRecordsAptamersLicence recorded
AptaNexus5,9465,943unknown (no licence terms found)
UTexas1,3261,289CC BY 3.0 US
AptaDB921874academic/unclear
Ribocentre428417academic/unclear
APTABASE365338academic/unclear

A further 7,513 source records were left out when the knowledge base was built and are not shown: records without a readable base sequence, sequences outside the sequence scope, targets that are themselves DNA or RNA, and records dropped by the screen of each paper's title and abstract.

How a plot is drawn

The centre
What was asked for: a target with the names the sources use for it, an aptamer with its stored fold, a pathway, a cell or a paper.
Aptamers
One dot each, coral for DNA and blue for RNA, with a dark outline when modified. The distance from the centre is the smallest Kd any of its records gives for the target at hand, on rings one decade apart. Dots beyond the dotted ring have no usable Kd stored: none was given, it had no unit, or it is flagged as suspect.
Spokes
For a target, one spoke per paper, in order of year; an aptamer sits on the spoke of the first paper that reports it. Papers without a stored year come last. For a pathway, a cell's proteins or a paper, one spoke per target.
Curved lines
Two aptamers where one sequence contains the other.
Numbered notes
Written from the numbers of the plot itself each time it is drawn.
What does not fit
A crowded plot draws the strongest binders of each spoke and says how many it left out. The table under the plot and its CSV always hold every record.
Folds
A drawing shows the fold the knowledge base stores. Nothing is folded when a page is drawn; only the positions of the bases on the page are worked out.

The five kinds of record

Aptamer
One base sequence with U and T as the source wrote them, DNA or RNA, modified or not. The same letters logged as DNA in one source and as RNA in another are two aptamers; so are a sequence and its modified copy. They are separate dots.
Source record
One record of one source database: an aptamer, a target, at most one measurement, the buffer, the paper. Every binding drawn on this site is one such record, and its page shows where the row sits in the source's file.
Target name
What the aptamer binds, as the source named it. Names are never merged: "Thrombin", "α-thrombin" and "Thrombin, Human" are three target names, because the source's wording is the authority.
Entity
One thing the names refer to: a protein, a compound, an organism, a cell line. A name links to an entity only when its identity was settled as one thing. Names that are ambiguous, mixtures, or mean different things in different papers have no entity. Pathways and relations hang on entities.
Pathway
A Reactome pathway whose reactions involve a protein entity directly, reached through the entity's UniProt entries.

Where a statement comes from

BasisMeaning
reportedA source states it in one cell, and the row quotes that cell.
inferredA rule reads it from indirect text, such as a paper's title. Low confidence.
calculatedComputed from the data: one sequence containing another, a predicted fold, a shared pathway, an expression link.
reported+calculatedBoth, for the same pair of aptamers.
not recordedNo source states it. This is not a negative.

Unknown is not no.

An aptamer with no recorded origin was not shown to be unselected; no source says how it arose.

An aptamer with no negative results was not shown to be specific; none were tested or none were reported.

An aptamer with no named parent is not thereby an original selection product; no source names one.

"N/A" and "not reported" in a source are missing values here, never "no binding".

Where it stops

Sequence families are a rough grouping
A family links a sequence of at least 15 nt to the sequences up to 40 nt longer that contain it. One short aptamer built into many sensor constructs pulls them all into one family, a shared primer can link unrelated clones, and a core inside a much longer clone is missed. Use families for browsing, not as curated families.
Calculated sequence links have no direction
That one sequence contains another does not say which was derived from which.
Calculated structures are of the bases alone
Folded with ViennaRNA at 37 °C, with DNA parameters for DNA and RNA parameters for RNA. Modifications are not modelled and G-quadruplexes are not predicted; a G-quadruplex former such as the 15-mer thrombin aptamer is therefore drawn without pairs.
Published structures are kept as published
Where a source gives a secondary structure without saying how it was folded, it is shown as given, with that caveat.
Stored affinities follow fixed conventions
A range is stored as the geometric mean of its ends. A value the source gives only as "greater than" is kept as a lower bound and left out of "at most" filters. Values that look wrong against the source text are flagged, not removed.

Identifier resources

ResourceRead asRelease or datesRetrievedTarget names it identifies or links
CLdump2026-06-082026-10-1814
Cellosaurusdump56.02026-10-18114
ChEBIdump2562026-10-18330
ComplexPortaldump2026-01-092026-10-181
GOdump2026-07-262026-10-1819
GlyTouCanweb service2026-10-186
MONDOdump2026-10-062026-10-18101
NCBIGeneweb service2026-10-18, 2026-10-19, 2026-10-20, 2026-10-24171
NCBITaxondump2026-10-182026-10-18605
PROdump73.12026-10-18468
PubChemweb service2026-10-18, 2026-11-01, 2026-11-04311
UniProtKBweb service2026_03342

Pathways: Reactome release 97. Expression links: Human Protein Atlas 25.1. Mouse cell type markers: PanglaoDB 2020-03-27.

This build

Knowledge base version 0.9.0; database file dated 2026-11-11. 115 of 115 build checks pass.

The build checks
CheckExpectedObservedPasses
binding rows == master rows89868986yes
evidence rows == master rows (no excluded record in the KB)89868986yes
evidence per source == master{'APTABASE': 365, 'AptaDB': 921, 'AptaNexus': 5946, 'Ribocentre': 428, 'UTexas': 1326}{'APTABASE': 365, 'AptaDB': 921, 'AptaNexus': 5946, 'Ribocentre': 428, 'UTexas': 1326}yes
step 01 records == master + excluded1649916499yes
step 01 records per source == master + excluded per source{'APTABASE': 610, 'AptaDB': 1350, 'AptaNexus': 12535, 'Ribocentre': 509, 'UTexas': 1495}{'APTABASE': 610, 'AptaDB': 1350, 'AptaNexus': 12535, 'Ribocentre': 509, 'UTexas': 1495}yes
each evidence row backs exactly one binding00yes
every aptamer has a binding00yes
master_row covers the master's rows once89868986yes
bindings per source == master{'APTABASE': 365, 'AptaDB': 921, 'AptaNexus': 5946, 'Ribocentre': 428, 'UTexas': 1326}{'APTABASE': 365, 'AptaDB': 921, 'AptaNexus': 5946, 'Ribocentre': 428, 'UTexas': 1326}yes
aptamer rows == master distinct seq_id78217821yes
aptamer_id 1:1 with seq_id78217821yes
distinct canonical sequences == master77737773yes
records per chemistry == master{'DNA': 6761, 'RNA': 2225}{'DNA': 6761, 'RNA': 2225}yes
records per target_class == master{'bacterium': 683, 'cell': 599, 'other': 111, 'parasite': 86, 'peptide': 103, 'protein': 4306, 'small_molecule': 1717, 'tissue': 6, 'toxin': 364, 'virus': 1011}{'bacterium': 683, 'cell': 599, 'other': 111, 'parasite': 86, 'peptide': 103, 'protein': 4306, 'small_molecule': 1717, 'tissue': 6, 'toxin': 364, 'virus': 1011}yes
is_modified records == master184184yes
measurements with a value == master rows with kd_nM_norm50235023yes
sum of stored Kd values == master (nM, rounded)325201963.897325201963.897yes
alias record counts sum to master rows89868986yes
target record counts sum to master rows89868986yes
per record seq_id, cluster, base and canonical sequence, source key, target_raw, seq_raw, seq_status == master00yes
one base sequence per aptamer (the key keeps U and T as written)00yes
mixed_letters exactly for a T in an RNA or a U in a DNA00yes
base_pos points at each base's letter in seq_raw (bases off)00yes
seq_suspect exactly when seq_status is unresolved_notation00yes
binding seq_status within step 02's statuses00yes
no binding has a nucleic-acid target (out of scope, step 04c)00yes
every binding is in sequence scope (step 02b)00yes
no stored Kd below zero00yes
lineage components == step 05 seq_cluster_id partition (in-scope sequences)[7181, 7181][7181, 7181]yes
calculated lineage edges only between in-scope aptamers00yes
clustering scope per record == master (in scope; exclusion reasons)[8986, {}][8986, {}]yes
in scope <=> seq_cluster_id set <=> no exclusion reason00yes
orphans binding.evidence_id -> evidence00yes
orphans binding.aptamer_id -> aptamer00yes
orphans binding.target_id -> target00yes
orphans binding.condition_id -> condition00yes
orphans measurement.binding_id -> binding00yes
orphans measurement.evidence_id -> evidence00yes
orphans measurement.condition_id -> condition00yes
orphans target_alias.target_id -> target00yes
orphans target_xref.target_id -> target00yes
orphans target_source_id.target_id -> target00yes
orphans target.entity_id -> target_entity00yes
orphans entity_relation.subject_id -> target_entity00yes
orphans entity_relation.object_id -> target_entity00yes
orphans entity_target_relation.subject_id -> target_entity00yes
orphans entity_target_relation.object_id -> target00yes
orphans entity_pathway.entity_id -> target_entity00yes
orphans entity_pathway.pathway_id -> pathway00yes
orphans origin.aptamer_id -> aptamer00yes
orphans origin.evidence_id -> evidence00yes
orphans lineage.parent_id -> aptamer00yes
orphans lineage.child_id -> aptamer00yes
orphans lineage_report.lineage_id -> lineage00yes
orphans lineage_report.parent_id -> aptamer00yes
orphans lineage_report.child_id -> aptamer00yes
orphans lineage_report.evidence_id -> evidence00yes
orphans lineage_unresolved.child_id -> aptamer00yes
orphans lineage_unresolved.evidence_id -> evidence00yes
orphans structure.aptamer_id -> aptamer00yes
orphans structure.evidence_id -> evidence00yes
orphans quality.evidence_id -> evidence00yes
orphans specificity.binding_id -> binding00yes
orphans specificity.aptamer_id -> aptamer00yes
orphans specificity.tested_target_id -> target00yes
orphans specificity.evidence_id -> evidence00yes
orphans evidence_note.evidence_id -> evidence00yes
orphans application.aptamer_id -> aptamer00yes
aptamers without an origin row00yes
origin methods within enum00yes
bindings without a target00yes
origin basis within enum00yes
origin 'unknown' exactly when basis is not_recorded00yes
inferred origin rows come only from the title rule00yes
reported origin rows quote a source cell00yes
lineage basis within enum00yes
reported lineage edges have a report, calculated-only edges none00yes
lineage_report rows == sum of n_reports11yes
lineage direction: reported edges 'reported', calculated-only 'unknown'00yes
specificity outcomes within enum and basis reported00yes
evidence notes are reported00yes
structure basis within enum; reported rows cite evidence, calculated rows none and name their parameters00yes
foldable exactly when a record reads the sequence reliably (seq_status ok or labels_removed, no ambiguity codes)00yes
calculated structures: one fold for every foldable aptamer77567756yes
calculated structures only on foldable aptamers, each with its chemistry's own parameters00yes
calculated dot-brackets balanced and as long as their aptamer00yes
G4 scores set exactly on the foldable aptamers00yes
every accepted name has an entity unless marked no; no other name has one00yes
entity_link set exactly with entity_id (identity or part_of)00yes
names of one identity CURIE share one entity00yes
entity n_names, n_records == its linked names00yes
entity uniprot == the union of its names' UniProt entries00yes
entity relations within enum, never an entity with itself00yes
entity_pathway rows == the entities' UniProt entries joined to Reactome's lowest-level pathways14041404yes
entity_pathway only through a UniProt entry of its entity00yes
target.uniprot == its UniProt xrefs that name the target00yes
every binding's identity_status is one of its target's (or the target differs by record)00yes
pathway n_entities == its entity_pathway rows, none without an entity00yes
shares_reactome_pathway == entity pairs with a common pathway, less has_species_entry pairs (subject < object, n_shared as counted)00yes
hpa_enriched_in == protein entities' HPA-enriched genes x cell entities' HPA lines of an enriched cancer type (rows in one set only)00yes
every cancer type HPA names for an enriched gene is a primary disease of HPA's cell line table[][]yes
every hpa_enriched_in row carries the line's nTPM00yes
HPA cell-type relations == protein entities' genes with a section's single-group label x cell entities whose CL term is the HPA cell type's, under or above it, with that match and step count (rows in one set only)00yes
entity_target_relation == protein entities' Cancer enriched genes x cell names without an entity, through their disease or a component line (rows in one set only)00yes
a PRO class reaches HPA through its human entry only if the curator confirmed it00yes
no HPA cell-type relation reaches a cell that states only non-human taxa00yes
panglaodb_marker_in == mouse protein entities' genes that are PanglaoDB mouse markers of a cell type meeting a mouse cell's CL term (rows in one set only)00yes
panglaodb_marker_in joins only mouse protein entities to mouse cell entities00yes
entity_target_relation: relations within enum, objects are names without an entity00yes
records mapped to a raw row whose sequence cell matches, per source{'APTABASE': 365, 'AptaDB': 921, 'AptaNexus': 5946, 'Ribocentre': 428, 'UTexas': 1326}{'APTABASE': 365, 'AptaDB': 921, 'AptaNexus': 5946, 'Ribocentre': 428, 'UTexas': 1326}yes
origin: every reported row equals its raw source cell (1485 rows)00yes
lineage_report: every reported row equals its raw source cell (1 rows)00yes
lineage_unresolved: every reported row equals its raw source cell (0 rows)00yes
specificity: every reported row equals its raw source cell (61 rows)00yes
evidence_note: every reported row equals its raw source cell (1005 rows)00yes
Tables and their row counts
TableRows
application0
aptamer7,821
binding8,986
condition2,386
entity_pathway1,404
entity_relation1,254
entity_target_relation45
evidence8,986
evidence_note1,005
kb_meta55
lineage859
lineage_report1
lineage_unresolved0
measurement6,198
origin8,103
pathway764
quality11,892
resource12
specificity61
structure13,780
target2,069
target_alias2,122
target_entity838
target_source_id878
target_xref2,761

Using it from a program

The API page lists the lookups behind the plots and lets you try them. Any id resolves at /id/<id>, for example /id/tgt:thrombin-72cda2. The full tables are under Downloads.