Protein Annotation · LiteFold
DisProt
Curated database of intrinsically disordered proteins and regions with experimental evidence and functional annotations.
What DisProt is
DisProt is a manually curated database containing intrinsically disordered proteins and disordered segments supported by experimental evidence. Intrinsically disordered proteins and regions do not adopt fixed three-dimensional structures under physiological conditions, yet they carry out diverse biological roles across cellular processes.
This distribution provides 3,199 protein entries encompassing 13,396 curated region annotations. Each record links protein sequences with experimental support, functional terms, taxonomy markers, and cross-references to external structural and domain databases.
How it was made
The dataset stores protein-level and region-level annotations in tabular Parquet format. Each protein entry incorporates the primary amino acid sequence, UniProt accession, organism, NCBI taxonomy identifier, overall disorder fraction, and coordinate boundaries for disordered segments alongside experimental evidence codes and literature references. Extended region-level metadata is also provided in a dedicated Parquet table.
Data partitioning into training and evaluation sets is deterministic, computed via sha256(disprot_id) % 10. Entries assigned to bucket 0 form the test split containing 324 proteins, while buckets 1 through 9 form the training split containing 2,875 proteins.
Key facts
- Total protein entries
- 3,199
- Curated region annotations
- 13,396
- Training split rows
- 2,875
- Test split rows
- 324
- Primary format
- Parquet
- Partitioning method
- Deterministic hash: sha256(disprot_id) % 10
- Publisher
- LiteFold
What you can use it for
- Disorder predictor training and benchmarking
- Train, validate, and benchmark sequence-based machine learning models on intrinsically disordered region detection using the fixed train and test splits.
- Disorder content profiling
- Filter and analyse proteins based on their overall disordered sequence proportion using the curated disorder_content metric.
- Experimental evidence validation
- Correlate structural predictions or physical measurements against experimentally verified disorder boundaries, evidence codes, and referenced literature.
Explore the data
The schema, real rows, full-text search and a SQL filter — read in your browser, nothing downloaded.
Explore
Read the real rows without downloading anything
Reading rows…
LiveRead from LiteFold/DisProt at the moment you asked. Nothing is cached or stored — every row above came from that request.
Strengths and limits
Good at
- Contains manually curated disordered regions backed by experimental evidence codes and published literature references.
- Provides a deterministic, reproducible split strategy across all entries.
- Includes cross-references to external biological resources such as UniProt, NCBI taxonomy, Pfam, Gene3D, and the Protein Data Bank.
Watch out for
- The dataset is limited in scale to 3,199 experimentally annotated protein entries.
- The partition is based on hash values of DisProt identifiers rather than sequence similarity clustering.
Questions
Can I look inside DisProt without downloading it?
Yes. This page reads its schema, real rows and per-column statistics in the browser, with full-text search and a SQL filter. Nothing is downloaded to your machine.
What licence is DisProt under?
Its authors declare cc-by-4.0. Read the terms on the original page before reusing the data.
How are the training and test splits separated?
Splits are determined using the hash calculation sha256(disprot_id) % 10. Bucket 0 is designated as the test set (324 entries), whereas buckets 1 to 9 constitute the training set (2,875 entries).
What information is provided for individual disordered regions?
Entries specify region start and end positions, region terms, curated region IDs, experimental evidence codes, literature citation IDs, and external database references.
Where can the standalone region annotations be found?
Region-level records are stored as an individual table accessible via metadata/regions.parquet.
How to cite it
Quaglia, Federica and Mészáros, Bálint and Salladini, Elisa and others. (2022). DisProt in 2022: improved quality and accessibility of protein intrinsic disorder annotation. Nucleic Acids Research, 50(D1), D480–D487. doi: 10.1093/nar/gkab1082
Written from DisProt's own dataset card. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.