# Computational Pipeline for Targeted Integration and Variable Payload Expression in Bacteriophage Engineering

## TL;DR

This paper introduces a computational pipeline to improve bacteriophage engineering by predicting optimal sites for inserting genetic payloads, validated through bioluminescent reporter experiments.

## Contribution

A novel computational pipeline for targeted integration of genetic payloads in phage genomes, using machine learning predictions and validated experimentally.

## Key findings

- A method was developed to identify optimal intergenic loci for payload insertion using PhagePromoter predictions.
- Three recombinant phages with reporter genes at distinct loci showed expression levels matching computational predictions.
- The pipeline enables temporal control of gene expression corresponding to early, middle, or late gene clusters.

## Abstract

Bacteriophages offer a promising alternative to conventional
antimicrobials,
especially when such treatments fail. While natural phages are viable
for therapy, advances in synthetic biology allow precise genome modifications
to enhance their therapeutic potential. One approach involves inserting
antimicrobial genetic payloads into the phage genome. These are typically
placed behind late-expressed genes, such as the major capsid gene
(cps). However, phages engineered with toxic payloads
often fail to produce viable progeny due to premature host shutdown.
To broaden the scope of viable genetic insertion sites, we developed
a method to identify intergenic loci with favorable expression profiles
using the machine learning-based promoter prediction tool, PhagePromoter.
Guided by these predictions, we designed a computationally assisted
engineering pipeline for targeted genomic payload integration. We
validated this approach by engineering bioluminescent reporter genes
into the genome of the strictly lytic Staphylococcus phage K at various predicted loci. Using homologous recombination,
we generated three recombinant phages, each carrying the reporter
at a distinct genomic location. These engineered phages exhibited
expression levels consistent with computational predictions and demonstrated
temporal expression patterns corresponding to early, middle, or late
gene clusters. Our study highlights the power of combining computational
tools with classical genome analysis to streamline phage engineering.
This method supports rational design and enables high-throughput,
automated phage modification, advancing the development of personalized
phage therapy.