Step 4: Mapping RXNs to Gene Families

Each per-species *_RXNS.txt file from Step 3 is mapped to UniRef90 gene families (GFs) using a hierarchical three-strategy approach.

Mapping Strategy Overview

Strategies are applied from most to least specific. Reactions unmapped by one strategy are passed to the next.

All RXNs with UniProtKB Accession  → GF assigned (UniProtKB accession pool)
         ↓
All RXNs → EC mapping
   EC hit? → GF assigned (EC + UniProtKB accession pooled if UniProtKB accession exists for that RXN)  → File 1 reactions
   No EC hit? → residual
         ↓
Residuals → Pfam mapping
   Pfam hit? → GF assigned (Pfam + UniProtKB accession pooled if UniProtKB accession exists for that RXN) → File 2 additions
   No Pfam hit? → final residual
         ↓
Final residuals → PA-only GF (if PA existed) or no GF → File 3 unique additions

Exact Pfam Set Matching — Detail

Each reaction can carry one or more Pfam domain sets (reflecting alternative subunit or isoform compositions). A GF is assigned only when its Pfam composition is an exact set match to the query.

Scenario	Pfam sets on reaction	Match result	GF assignment
Single set, full match	{PF00001}	{PF00001} found	✅ Assigned
Single set, partial match	{PF00001, PF00002}	Only {PF00001} found	❌ Not assigned
Single set, no match	{PF00001}	Not in database	❌ Not assigned
Multiple sets, all match	{PF00001}, {PF00002, PF00003}	Both found	✅ GFs from both sets pooled
Multiple sets, partial match	{PF00001}, {PF00099}	Only first found	✅ GFs from matched set only

Running the Mapping

The mapping runs as a SLURM array job — one task per BioCyc organism file.

Before submitting, set these variables at the top of scripts/map_RXNS2GF.sh:

Variable	Description
BIOCYC_RXNS	Directory containing per-organism *_RXNS.txt files from Step 3
BASE_DIR	Output root directory (subdirectories created automatically)
SCRIPT_DIR	Directory containing the Python scripts
DB	Path to uniref90_pf_ec.db built in Step 2

Array size: Set --array=0-N where N = (number of files) - 1. For clusters with a MaxArraySize limit, submit in batches using the OFFSET variable:

# Standard submission
sbatch --array=0-N scripts/map_RXNS2GF.sh

# Batched submission example
OFFSET=6000 sbatch --array=0-5999 scripts/map_RXNS2GF.sh

Each task runs all mapping steps sequentially with checkpointing, so incomplete runs can be safely resumed.

Output

Three directories are created under BASE_DIR:

Directory	Contents
rxn_ec_uniref90/	EC-number based RXN→GF mappings
rxn_exactpfamset_uniref90/	Exact Pfam set RXN→GF mappings
rxn_proteinaccession_uniref90/	UniProtKB Accession based RXN→GF mappings

Each directory contains:

• RXN_GF-{NAME}.txt — reactions successfully mapped to a GF
• RXN_NO_GF-{NAME}.txt — unmapped residuals passed to the next strategy

---

Next: Step 5 — Aggregating Multi-Species Mappings