The KMT2A (MLL) gene rearrangements (KMT2A-r) are associated with a diverse spectrum of acute leukemias. Although most KMT2A-r are restricted to nine partner genes, we have recently revealed that KMT2A-USP2 fusions are often missed during FISH screening of these genetic alterations. Therefore, complementary methods are important for appropriate detection of any KMT2A-r. Here we use a machine learning model to unravel the most appropriate markers for prediction of KMT2A-r in various types of acute leukemia.
Batch_effect.R & Analysis_correlation.R
First, we integrate the TARGET, Beat AML and EGA datasets and then remove the batch effect using the study variable (dataset origin). After that, we remove the variables with high correlation (> 90%).
KMT2Ar-preProcess&Boruta.ipynb
Variance Analysis and feature selection with BorutaPy. 247 genes were considered important for predicting the KMT2A gene rearrangement.
KMT2Ar-Development-247genes.py
Development of the models using 247 genes selected by BorutaPy.
KMT2Ar-Model_Analysis-247Genes.ipynb
Analysis of the developed models (confusion matrix; metrics; ROC curve; Shap values).
The RData file - Dataset_ML_KMT2Ar.RData - available only contains the TARGET and Beat-AML data due to the control of the others (EGA studies).
Downloading and obtaining data are described in the materials and methods section of the paper.
Please cite: Lopes BA, Poubel CP, Teixeira CE, Caye-Eude A, Cavé H, Meyer C, Marschalek R, Boroni M and Emerenciano M (2022) Novel Diagnostic and Therapeutic Options for KMT2A-Rearranged Acute Leukemias. Front. Pharmacol. 13:749472. doi: (https://doi.org/10.3389/fphar.2022.749472).
Bioinformatics and Computational Biology Laboratory (LBBC-INCA); Brazilian National Cancer Institute (INCA-RJ) | LBBC team (https://sites.google.com/view/bioinformaticainca-en/home-en)