Pulse Brain · Growing Health Evidence Index
Tier 3 — Observational / field trialPeer-reviewed

Benchmarking of deep neural networks for predicting personal gene expression from DNA sequence highlights shortcomings

Alexander Sasse, Bernard Ng, Anna Spiro, Shinya Tasaki, David A. Bennett, Chris Gaiteri, Philip L. De Jager, Maria Chikina, Sara Mostafavi

Nature Genetics · 2023

Read source ↗ All evidence

Summary

This benchmarking study (2023) evaluates deep neural network approaches for predicting individual gene expression profiles directly from DNA sequence, as published in Nature Genetics. The authors systematically assess model performance and identify substantial limitations in current methods' ability to generalise across populations and datasets. The work suggests that sequence-based gene expression prediction remains a challenging problem with important methodological gaps that warrant further investigation.

Regional applicability

This is a fundamental computational biology study with no direct application to United Kingdom farming systems, soil health, or agricultural practices. Its relevance to Vitagri's Pulse Brain catalogue is marginal unless intended to support future work on genomic approaches to crop or livestock trait prediction.

Key measures

Deep neural network model performance metrics for gene expression prediction; cross-dataset generalisation; comparison of model architectures and training approaches

Outcomes reported

The study benchmarked deep neural network models for predicting personal gene expression directly from DNA sequence data and evaluated their predictive performance and generalisation across datasets. The work highlights significant shortcomings in current approaches to sequence-based gene expression prediction.

Theme
Measurement & metrics
Subject
Other / interdisciplinary
Study type
Research
Study design
Benchmarking study / Methodology evaluation
Source type
Peer-reviewed study
Status
Published
Geography
United States
System type
Laboratory / in vitro
DOI
10.1038/s41588-023-01524-6
Catalogue ID
SNmp6e739j-923bkw

Topic tags

Pulse AI · ask about this record

Dig deeper with Pulse AI.

Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.