# Machine learning meets psoriasis: identifying key lactylation biomarkers as potential targets for diagnosis and therapies

## TL;DR

This study uses machine learning to identify lactylation-related biomarkers for psoriasis, offering new diagnostic and therapeutic targets.

## Contribution

The paper introduces novel lactylation-related biomarkers for psoriasis using machine learning and experimental validation.

## Key findings

- MPHOSPH6, ENO1, MKI67, and FABP5 are identified as key lactylation-related biomarkers for psoriasis.
- High levels of MPHOSPH6 and ENO1 are risk factors for psoriasis according to Mendelian randomization analysis.
- 103 potential drugs targeting these biomarkers were identified from the DSigDB database.

## Abstract

Psoriasis is a long-term autoimmune skin condition marked by repeated inflammation. Recent findings indicate that affected skin in psoriasis shows increased aerobic glycolysis and lactate buildup, suggesting that protein lactylation may play a role in the disease. However, biomarkers related to lactylation for diagnosing and treating psoriasis remain poorly defined.

Initially, genes with altered expression in psoriasis were identified. Key gene modules from Weighted Gene Co-expression Network Analysis (WGCNA) were used to pinpoint psoriasis-associated genes. These genes were then intersected with lactylation-related genes. Random Forest and LASSO regression algorithms selected lactylation-related biomarkers. Mouse psoriasis models were created using imiquimod to validate key gene expression. The immune microenvironment in psoriasis lesions was analyzed with CIBERSORT. Regulatory networks of miRNAs(microRNAs)-genes and TFs(Transcription Factors)-genes were built using NetworkAnalyst. Potential drugs targeting these biomarkers were predicted via the DSigDB database, and their expression and distribution were visualized in single-cell sequencing data. Finally, two-sample Mendelian randomization and summary data-based Mendelian randomization were performed to investigate the causal relationship between the biomarkers and psoriasis.

A total of 1,623 key genes associated with psoriasis were identified through differential gene screening and WGCNA analysis. Among these, 26 were related to lactylation. Machine learning pinpointed MPHOSPH6, ENO1, MKI67, and FABP5 as lactylation-related biomarkers for psoriasis, with ROC curves confirming their strong diagnostic capabilities. RT-qPCR experiments validated their reliability, and immune infiltration analysis showed significant correlations with immune cells. Additionally, 103 drugs targeting these biomarkers were found in the DSigDB database. Mendelian randomization analysis suggested that high levels of MPHOSPH6 and ENO1 are risk factors for psoriasis.

MPHOSPH6, ENO1, MKI67, and FABP5 are identified as lactylation-related biomarkers for psoriasis, with MPHOSPH6 and ENO1 overexpression posing as risk factors. These findings offer potential new diagnostic and therapeutic targets for the disease.