Identification of human antibodies using bottom-up proteomics approaches relies on database searches that compare experimental peptide values and their fragments to theoretical values derived from protein sequences in a database. Conventional protein databases like UniProt contain only a fraction of the antibody sequences present in the human body, with just 1,095 reviewed sequences and 37,848 unreviewed sequences as of May 2024, despite the human body's capacity to produce quintillions of antibodies. This limitation underscores the urgent need for new solutions to identify expressed antibodies.

We propose using antibody sequence databases based on previous genomic studies to address this limitation. Leveraging billions of human antibody sequences from 70 genomic studies (contained in the Observed Antibody Space database: https://opig.stats.ox.ac.uk/webapps/oas/), we created disease-specific antibody databases for bottom-up proteomics to identify new antibodies associated with relevant diseases such as SARS-CoV-2, HIV, Ebola, and chronic lymphocytic leukaemia. Instead of directly using antibody sequences, which have considerable redundancy due to antibody constant regions, we used peptide sequences corresponding only to the variable regions of antibodies. We performed in silico digestion of antibody sequences to obtain peptide sequences and selected peptides in the variable regions of antibodies. The peptides already present in the two major human protein databases, UniProt and NCBI-RefSeq, were filtered out. Subsequently, we merged the resultant peptides with the UniProt human proteome and the common Repository of Adventitious Proteins (cRAP) to form disease-specific peptide databases for bottom-up proteomics.

We demonstrated the use of these databases in identifying new SARS-CoV-2 antibody peptides in previously measured blood samples, validated by running the same analysis on negative control samples. Additionally, we created an online resource for researchers to download disease-specific peptide databases in fasta format. This study provides researchers with antibody resources comprising millions of antibodies for each disease, ready to be used as part of common proteomics data analysis workflows. Our databases will aid researchers in discovering novel disease-specific antibodies in mass spectrometry data, adding a new layer of valuable information, not provided when using commonly used databases such as UniProtKB and NCBI-RefSeq.