# Memory-efficient, accelerated protein interaction inference with blocked, multi-GPU D-SCRIPT

## TL;DR

This paper introduces a memory-efficient and accelerated version of D-SCRIPT for predicting protein interactions at scale using multiple GPUs.

## Contribution

The novel contribution is blocked multi-GPU parallel inference, which significantly reduces memory usage and enables large-scale PPI analysis.

## Key findings

- Blocked multi-GPU inference reduces memory usage by 13.8× for large proteomes.
- The new method enables multi-GPU parallelism for PPI inference.
- The updated D-SCRIPT is publicly available with these improvements.

## Abstract

D-SCRIPT is a powerful tool for high-throughput inference of protein–protein interactions (PPIs), but it is expensive in time and memory to infer all PPIs for network-/proteome-level analyses. We introduce D-SCRIPT with blocked multi-GPU parallel inference, which substantially reduces memory usage across tasks and computational systems (13.8× for a representative large proteome) and enables multi-GPU parallelism.

Blocked multi-GPU parallel inference has been integrated into the main D-SCRIPT package, available at https://github.com/samsledje/D-SCRIPT. An archived version of the code at time of submission can be found at https://doi.org/10.5281/zenodo.16325182.