2023 World Conference on Lung Cancer (Posters)-EP02.02. Machine Learning for Multi-Omics Data Identifies Vulnerabilities in a Subset of Squamous Cell Lung Cancers

EP02.02. Machine Learning for Multi-Omics Data Identifies Vulnerabilities in a Subset of Squamous Cell Lung Cancers - PDF(Abstract)

Back to course

Pdf Summary

In this study, researchers used a machine-learning-based computational pipeline to identify potential therapeutic targets for squamous cell lung cancer (LUSC), a common subtype of non-small cell lung cancer. They focused on the lineage-survival oncogene SOX2, which is dysregulated in a large proportion of LUSC cases. Despite the prevalence of LUSC, there are currently no targeted therapies approved for this type of cancer.<br /><br />The researchers developed a pipeline to integrate and analyze large-scale multi-omics datasets from public databases and a custom SOX2-dependent model of early LUSC. They conducted a network analysis to identify drug target candidates and then evaluated these targets in cell-based models using small molecule inhibitors, siRNA, and CRISPR-Cas9.<br /><br />Using their pipeline, the researchers identified the SOX2 interactome in LUSC and identified candidate drug targets. They then scored and ranked these targets based on their druggability and their connection to the SOX2 network. The top 10 candidates were selected for experimental validation.<br /><br />The results of the validation study showed that existing small molecule inhibitors targeted the PI3K/AKT/mTOR pathway, which was validated as a valid drug target in an organotypic SOX2-dependent model of early LUSC. Additionally, two transcription factors, KLF5 and FOXM1, were validated as targets in a subset of LUSC.<br /><br />The researchers concluded that their machine-learning-based pipeline successfully identified new druggable pathways and candidate targets for LUSC. They confirmed the therapeutic potential of these targets in vitro and believe that this approach could help meet the unmet needs in LUSC and other cancer types.<br /><br />Overall, this study highlights the use of machine learning and computational analysis to identify potential therapeutic targets for a specific subtype of lung cancer. The pipeline developed in this study could potentially be applied to other cancer types to identify new treatment options.

Asset Subtitle

Daniel Kottmann

Meta Tag

Speaker Daniel Kottmann

Topic Tumor Biology: Preclinical Biology - Omics Approaches

Keywords

therapeutic targets

squamous cell lung cancer

LUSC

non-small cell lung cancer

lineage-survival oncogene SOX2

targeted therapies

multi-omics datasets

drug target candidates

small molecule inhibitors

transcription factors