Protein database cleanup

Process and clean up protein databases for proteomics experiments by selecting the longest or most prevalent isoforms from protein databases, minimising overlap in proteins and avoiding multiple mappings for peptides.

The main script

ProteinDatabaseCleanup.sh works by linearizing the sequences and removing sequences without GN IDs and uncharacterized and fragmented sequences.

1. The resulting sequences are unique by gene names (identified by the pattern following GN=).
2. If there are multiple sequences with the same gene name, the following priority order should be used to select one: 1) The sequence from the SwissProt database is preferred. 2) If multiple sequences from SwissProt exist, the longest should be chosen. 3) If no sequence from SwissProt exists, but there are sequences from TrEMBL, the longest should be chosen.

Modifications of the script can include maintaining a certain level of sequence diversity in the dataset and keeping sequences that are less than 90% similar: t_coffee -other_pg seq_reformat -in inter_out_deleted_empty_lines -action +trim seq%%90_ >> deduped.fasta

However, our goal is to minimize redundancy and keep only unique representative sequences per gene, so we are selecting the longest sequence per gene as per the database priority (SwissProt over TrEMBL).

How to use

1. Download the latest Uniprot entries for the selected organism in fasta (gz-archive, canonical only).
2. Clone this repository to your local machine.
3. Ensure that you have Bash installed and that the scripts have execute permissions (for example, chmod +x ProteinDatabaseCleanup.sh gives the script execute permissions).
4. Run the ProteinDatabaseCleanup.sh script with your protein database as an argument.

Requirements

1. Bash 4.0 (or later)
2. AWK
3. t_coffee (for sequence alignment)

This project is open-source, and contributions are welcomed. Please feel free to open an issue or submit a pull request if you have suggestions for improvements or have found any issues.