LPMD Algorithm: Methylation Disorder Data for Predicting Azacitidine Response in AML

68 AML patients from the GSE62298 dataset were used to develop the Local Promoter Methylation Disorder algorithm. The study validated findings in a 20-sample WGBS cohort and a 194-sample TCGA-LAML cohort, and assessed clinical predictive value in 63 high-risk MDS/secondary AML patients. The work by Xin Liu, last updated in March 2026, identifies a 5-DMDR panel for treatment response prediction.

Use Cases

• Predicting azacitidine treatment response based on a 5-DMDR panel derived from multi-algorithm consensus.
• Analyzing methylation instability patterns associated with DNMT3A mutations using a 1-kb sliding window algorithm.
• Monitoring treatment dynamics by tracking changes in Local Promoter Methylation Disorder scores over time.
• Identifying differentially methylated disorder regions (DMDRs) exhibiting bidirectional patterns of aberrant stabilization or maintenance failure.

Strengths

• Algorithm development and validation involved multiple cohorts, including 68, 20, and 194 patient samples.
• Clinical predictive panel achieved an AUC of 0.777 with 81% sensitivity and 73% specificity.
• Longitudinal analysis included 153 samples for treatment dynamics assessment.

Limitations

• Dataset is a 44.0 KB PDF; the underlying raw methylation data is not directly provided.
• Column-level documentation is absent; field semantics must be inferred from the publication.
• Row count for the primary data matrices is unknown, which may limit suitability assessment.

Provenance

Source

figshare, authored by Xin Liu.

Collection Method

Algorithm developed using the GSE62298 dataset and validated in independent WGBS and TCGA-LAML cohorts.

Freshness

Last updated 2026-03-18 05:21:26