Predicted pockets
Geometry-based predictions on AlphaFold DB v6 models. The score is a method output, not validated confidence. Pocket pLDDT and PAE are unavailable.
About MemPoc & user guide
Understand the record.
Follow the evidence.
MemPoc brings membrane-protein annotations, small molecules, relationship evidence, predicted pockets, and disease associations into one research workspace.
Open the explorerThe database
Connected records. Preserved context.
Start with a protein or compound, then inspect the evidence connecting them. Protein records also bring together source annotations, predicted structural pockets, and gene–disease context. Each view keeps source provenance and quality flags close to the data.
Records
Proteins and compound identifiers are the entry points. Registered compounds include structural ligands and other molecules, as well as drugs.
Relationships
A protein–compound pair may have several evidence contexts. A context count is different from a count of distinct pairs.
Provenance
Inspect the source, evidence type, scope, directness, and QC status before interpreting a relationship.
Search & refine
One search, every record type.
Type a gene, protein, compound, UniProt accession or InChIKey. One search covers proteins, compounds and evidence contexts; the record-type tabs filter the results. Exact identifiers and names are listed first, and suggestions appear as you type.
| Write | To find | Example |
|---|---|---|
words | Every word must match, in any order | insulin receptor |
"phrase" | Words in this exact order | "insulin receptor" |
OR | Either side may match | gefitinib OR erlotinib |
NOT or -word | Exclude records that match | kinase NOT receptor |
( ) | Group conditions; NOT binds before AND, AND before OR | (EGFR OR ERBB2) AND kinase |
What each record type searches
Proteins: UniProt accession, gene symbol, protein name and NCBI Gene ID. Compounds: identifier, name and InChIKey. Evidence contexts: the protein and compound text of each context together, so EGFR gefitinib finds contexts that link both. Words match the start of words: recep finds “receptor”, but GFR does not find EGFR. In Evidence contexts, exact protein and compound filters and the evidence type also apply, combined with AND.
Read the evidence
Different evidence answers different questions.
Relationship contexts can represent different kinds of source support. They are not uniformly direct binding measurements or independent experiments.
- Assay activity
- Measurements recorded in an assay context. Read the endpoint, comparator, unit, and QC status together.
- Curated target assertion
- A source-reported target relationship. Inspect its scope and directness before treating it as binding evidence.
- Structure–ligand observation
- A ligand observed in a structural context. Its biological role and interpretation matter.
- UniProt binding-site annotation
- A source annotation of a binding site. Follow the record’s provenance for its supporting context.
Project evidence categories
Disease labels such as medium, high, and very high describe the project’s evidence policy. They do not establish causality or a disease-risk probability.
An empty search result means no matching record was found in this snapshot. It does not establish absence of an interaction, binding site, or disease association.
Release & limitations
Know the scope of your analysis.
This portal uses the v3.1 collaborator snapshot. It includes protein sequence lengths; full sequences, FORMAL binding-evidence tiers, negative evidence, and v7.2.1 overlays are not included.
Make your analysis traceable.
Keep the release identifier, source commit, and download checksum with your results. The data page includes all three.