Protein–Protein Interaction Database Design and Query System

A fully normalized relational database for storing, querying, and analyzing protein–protein interaction data from the Protein–Protein Docking Benchmark v5.5.

Final Database Statistics

Metric	Value
Complexes	257
Proteins	489
Proteins with Sequences	489 (100% ✓)
Unique PDB Structures	472
Average Sequence Length	1,439 residues
Total Records	1,479+

Project Goals

This database enables researchers to:

• ✅ Query complexes by interaction type (AA, EI, ES, ER, OG, OR, OX, AS)
• ✅ Retrieve protein sequences and structural information
• ✅ Filter complexes by docking difficulty (Rigid/Medium/Difficult)
• ✅ Identify proteins that participate in multiple complexes
• ✅ Analyze RMSD and interface surface area measurements

Docking Difficulty Distribution

Difficulty	Number of Complexes	Percentage
Rigid	162	63.0%
Medium	60	23.4%
Difficult	35	13.6%

Complex Type Distribution

Code	Description	Count	%
AA	Antibody-Antigen	55	21.4%
OX	Others, miscellaneous	55	21.4%
EI	Enzyme-Inhibitor	45	17.5%
ER	Enzyme with regulatory/accessory	26	10.1%
OR	Others, Receptor containing	24	9.3%
OG	Others, G-protein containing	23	8.9%
ES	Enzyme-Substrate	17	6.6%
AS	Antigen-Single domain Antibody	12	4.7%

Proteins in Multiple Complexes (Top 5)

Protein	Complexes	PDB ID
Rac GTPase	5	1MH1
Actin	4	1IJJ
p300 TAZ2 domain	3	3IO2
Ubiquitin	3	1YJ1
MEF2A	3	3KOV

️ Technologies Used

• Database: SQLite
• ETL & Analysis: R with packages:
  • DBI, RSQLite (database connection)
  • dplyr, tidyr (data manipulation)
  • bio3d (PDB sequence extraction)
  • ggplot2 (visualization)
  • readxl (Excel import)

Database Schema Design

The schema is fully normalized to Third Normal Form (3NF):

Core Tables

Table	Purpose	Records
complex	Protein-protein complexes	257
protein	Unique proteins with sequences	489
structure	Bound/unbound PDB records	733
complex_protein	Junction table (many-to-many)	489+
interaction_feature	RMSD, iSA, benchmark version	257
complex_type	Interaction categories (lookup)	8
difficulty	Docking difficulty levels (lookup)	3

ER Diagram

ETL Workflow

1. Extract

• Loaded Table_BM5.5.xlsx (Docking Benchmark v5.5)
• Parsed complex IDs, PDB IDs, chain IDs, and interaction metrics

2. Transform

• Cleaned and standardized controlled vocabularies
• Separated bound/unbound structures
• Removed duplicate records
• Added sequence data from PDB files

3. Load

Loaded into SQLite tables in dependency order:

1. Lookup tables (complex_type, difficulty)
2. structure (bound/unbound records)
3. protein (unique proteins)
4. complex (complexes with foreign keys)
5. complex_protein (junction table)
6. interaction_feature (measurements)

Sequence Addition

# Sequences were fetched from PDB files using bio3d package
# Successfully added sequences for all 489 proteins (100% coverage)