Critical Assessment of Protein Intrinsic Disorder (CAID)

This is the official repository of the Critical Assessment of Protein Intrinsic Disorder (CAID) challenge.

The previous CAID2 repository is available here.

Follow the instructions below to replicate the generation of the references and some useful statistics.

# Generate the folder structure
mkdir -p data/{disprot,sifts,alphafold,output/{references,references_stat,references_merge_analysis,homology,new_taxdump}}

references

Generate the references from two snapshots of the DisProt database (mongo export) DisProt data can be obtained directly exporting the relevant database collections (ask the developers). Or using the download service from the website (lastest annotations might not be available to the public). Note the formats are slightly different.

Download DisProt data

Use MongoDB compass and download the current public collection and the current "curators" collections.

• Public 2024_12 # public at the time predictors submitted their software
• Current 2025_06_c # 1 July 2025, 15:55 CEST

To download directly from MongoDB with a command (e.g. CAID3 CASP-16 dataset) mongoexport --uri "mongodb://moros:27017/disprot8" --collection entries_2025_06_c > data/disprot/disprot8.entries_2025_06_c.json

references_stat

Generate statistics about the references

homology

Parse the blast output, extract information about the best match and perform optimal pairwise alignments. Comparison are between CAID and DisProt "old" and between CAID and PDB seqres.

# Download SIFTS data to get the list of SEQRES sequences in the PDB 
wget -O data/sifts/uniprot_segments_observed.tsv.gz ftp://ftp.ebi.ac.uk/pub/databases/msd/sifts/flatfiles/tsv/uniprot_segments_observed.tsv.gz
wget -O data/disprot/go-basic.obo http://purl.obolibrary.org/obo/go/go-basic.obo

# Generate BLAST alignments of the new DisProt against the old DisProt and against PDB seqres
# Install blast on your home (check the version and paths)
wget https://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST/ncbi-blast-2.15.0+-x64-linux.tar.gz
tar -xf ncbi-blast-2.15.0+-x64-linux.tar.gz
export PATH="/home/$USER/ncbi-blast-2.15.0+/bin:$PATH" 

# Download PDB seqres
wget https://files.wwpdb.org/pub/pdb/derived_data/pdb_seqres.txt.gz -O data/output/homology/pdb_seqres.txt.gz
gunzip data/output/homology/pdb_seqres.txt.gz

# Make blast dbs
makeblastdb -in data/output/homology/disprot_old.fasta -dbtype prot
makeblastdb -in data/output/homology/pdb_seqres.txt -dbtype prot

# Run BLAST
blastp -db data/output/homology/disprot_old.fasta -query data/output/homology/disprot_new.fasta -out data/output/homology/disprot_new_old.blast -outfmt 6 -num_threads 12
blastp -db data/output/homology/pdb_seqres.txt -query data/output/homology/disprot_new.fasta -out data/output/homology/disprot_new_pdb.blast -outfmt 6 -num_threads 12

homology_plot

Generate plots from the output of the homology notebook

# Download taxonomy data
wget -O data/new_taxdump.tar.gz  ftp://ftp.ncbi.nih.gov/pub/taxonomy/new_taxdump/new_taxdump.tar.gz
tar -xf data/new_taxdump.tar.gz -C data/new_taxdump

references_merge

Combine predictions with reference data. It requires the predictions, the references, the mapping of predictions to the different challenges and the assessment (F1-score) one for each reference

reference_merge_analysis

Generate figures comparing predictions and references, one for each target