Hi everyone,
I'm a beginner in Omics. Now I'm working on early cancer screening and cancer prognosis using TCGA DNA methylation data. So far, I have learned and performed differential methylation analysis between normal and tumor groups. I identified differentially methylated probes with criteria.
However, I am confused about what I should do next to prioritize clinically relevant probes/biomarkers. I only know that I need to use this data for early screening and prognosis, but I know very little about it.
I have noticed that this post recommends an analytical approach What is the best way to combine machine learning algorithms for feature selection such as Variable importance in Random Forest with differential expression analysis?. After reading this, I plan to adopt a suggested workflow:
1.Perform PCA/clustering on differentially methylated probes (DMPs).
2.Build regression models to link methylation patterns with clinical outcomes.
I am not sure if my understanding is correct. And I have two questions about the workflow:
First, I don't know which algorithm or model to use? Do you have any recommendations?
Second, I am unsure which clinical information to use as covariates if I plan to build the regression model. The data I obtained from TCGA includes the following information: age at initial pathologic diagnosis, gender, race, ethnicity, vital status, stage event, histological type, neoplasm histologic grade, and residual tumor. Can these variables help me with early screening and prognosis?
Any tool recommendations or best practices would be greatly appreciated. Thank you in advance for your time and consideration!
i dont recommend continuing on this path until you can answer these questions easily for yourself
Dear LauferVA,
Thank you for your sincere advice. You are right, and I fully agree with you. I have just started to get involved in bioinformatics and I am still self-studying. As a student without a background in biology or medicine, I have been continually confronted with challenges related to ensuring biological plausibility. This has always made me lack confidence in my analysis. I will give priority to addressing the suggestions you provided. Thank you for your reply and guidance.