Digested Peptide Database

Overview

DigestedProteinDB is a database that enables researchers in the field of proteomics to search for enzymatically digested peptides by mass, sequence, and m/z values.

A Java library for creating and searching a digested peptide database with RocksDB key-value store. Digested peptides are generated from protein sequences using a specific enzyme (e.g., trypsin) and are stored in a database for rapid mass-based searches. This library is designed to be a compact and efficient storage of in-silico digested peptides.

The database is designed to be a compact and efficient storage of in-silico digested peptides. The primary goal is to support rapid mass-based searches, even on less powerful desktop machines.

https://digestedproteindb.pbf.hr

This compact database layout enables fast peptide mass queries with minimal resource usage. For example, a Uniprot TrEMBL bacteria dataset (7 to 30 amino acids, 1 missed cleavage) occupies about 50 GB on disk and supports quick lookups by mass range on standard hardware.

For example, searching for masses from: 1500.6 Da to 1500.8 Da gives JSON ptmSearchResults like:

acc — Uniprot accession number seq - peptide sequence, mass - peptide mass.

{
  "totalResult": 1543,
  "memory": "2016 MB",
  "duration": "00:00:00.031",
  "page": 1,
  "pageSize": 10,
  "result": [
    {
      "1500.6": [
        {
          "seq": "SSNESGSSGSMTNEK",
          "acc": [
            "A0A085MEL6"
          ]
        },
        {
          "seq": "CSGSTSESEGSQTNK",
          "acc": [
            "A0AAD6LB27",
            "A0AAD6L9P6",
            "A0A4U5R0T2",
            "A0A8X8BWN4",
            "A0A8X8C0W2"
          ]
        },
        {
          "seq": "NSCSASESQSDTGTK",
          "acc": [
            "A0A3P9Q6N9"
          ]
        },
        {
          "seq": "EDNSCSSSAGSTQTK",
          "acc": [
            "A0AAV4GC55"
          ]
        }
      ]
    },
    {
      "1500.6002": [
        {
          "seq": "VEIEYDDEDMMI",
          "acc": [
            "A0A1Y2LUM3"
          ]
        },
        {
          "seq": "CHGWGGSQCHHHR",
          "acc": [
            "A0A8B9NSG9"
          ]
        }
      ]
    },
    ... 

Key Features

Minimal and fast database: Database of digested peptides designed for a small on-disk footprint Mass-oriented search: Stores peptide data keyed by mass to expedite lookups based on precise mass ranges

Library Integration

This database is designed as a Java library that can be easily embedded into existing software applications. The library provides a clean API for programmatic access to peptide data, making it ideal for integration with mass spectrometry analysis tools or custom bioinformatics pipelines.

Installation

The project can be built using Maven and Java 24. To build the project, navigate to the root directory of the project and run:

mvn clean package

This will create a JAR file in the target directory.

Requirements

• Java 24 or higher
• Maven 3.8 or higher for building the project from source
• Linux, MacOS for database building
• Linux, MacOS, Windows for running the database search server

Command-Line Parameters

The application provides two main commands: create-db for creating the database and server for running the web server. Below is a detailed explanation of the parameters for each command.

create-db Command

This command is used to create a digested peptide database. The following table describes the available parameters:

Parameter	Description	Required	Default Value
-d, --db-dir	Path to the database directory.	Yes	None
-u, --uniprot-xml	Path to the Uniprot XML file (.xml.gz or .xml).	Yes	None
-n, --db-name	Name of the database.	Yes	None
-c, --clean	Clean all files in the database directory before creating.	No	false
-m, --min-length	Minimum peptide length.	No	7
-M, --max-length	Maximum peptide length.	No	30
-mc, --missed-cleavage	Number of missed cleavages.	No	1
-s, --sort-temp-dir	Path to the temporary directory used for sorting.	No	None
-ncbi, --ncbi-taxonomy-path	Path to the NCBI taxonomy file (nodes.dmp). https://ftp.ncbi.nlm.nih.gov/pub/taxonomy/taxdump.tar.gz	No	None
-p, --taxonomy-parents	NCBI Taxonomy parent (ancestor) IDs.	No	All
-e, --enzyme	Enzyme used for digestion (e.g., Trypsin, Chymotrypsin).	No	Trypsin
-t, --taxonomy-division	NCBI Taxonomy division to filter proteins (e.g., Bacteria, Viruses, etc.).	No	ALL

Example Usage

Run the compiled JAR with the create-db subcommand:

java -jar digestdb.jar create-db 
-d "path/to/db/dir" 
-u "path/to/uniprot/xml/file.xml.gz" 
-n myDatabaseName
-c 
-m 7 
-M 30 
-s path/to/sort/temp/dir 
-ncbi "path/to/ncbi/taxonomy/file" 
-p 10239 -p 33090 
-e Trypsin 
-t ALL

Run the Web Server

Start a simple web server for searching:

java -jar digestdb.jar server 
-p 8080 
-d "/path/to/db/dir"

Demonstration search web: https://digestedproteindb.pbf.hr

Create database steps

The database creation process starts with the uniprot.xml.gz (or .xml) file. Example class hr.pbf.digestdb.CreateDatabase is used to create the database.

The process consists of the following steps:

1. Parse Uniprot Data: Extract sequences and relevant metadata from uniprot .xml.gz database and NCBI Taxonomy.
2. Digest Proteins: Cleave the protein with a specific enzyme (trypsin, for example).
3. Organize Data: Sort and group peptides by mass for easy storage.
4. Build Database: Generate a mass-indexed key-value engine (RocksDB) and custom array list index for quick searches.

Many optimizations are used to speed up the process, including:

• Custom Data Structures: Use specialized data structures for storing data. Sequences are stored as 5-bit encoded strings, masses are stored as 4-byte integers, and accession numbers are 36-base encoded strings stored in 64-bit data (long) and mapped to 4-byte integers.

TODO

• Python code for searching the database
• Expand and improve documentation
• Create more downloadable databases with different enzymes and taxonomy units
• Develop native version for Windows, MacOS, and Linux without Java and RockDB dependencies

This project uses JProfiler, a powerful profiling tool, for performance analysis and optimization. We thank ej-technologies GmbH for providing an open-source license.