Consensus Biomarker Selection for Cancer Diagnosis
Author Information
Author(s): Janusz Dutkowski, Anna Gambin
Primary Institution: Institute of Informatics, Warsaw University
Hypothesis
Can a consensus approach to biomarker selection improve classification accuracy in cancer diagnosis using mass spectrometry data?
Conclusion
The proposed methodology can improve the classification results and provide a unified biomarker list for further biological examinations.
Supporting Evidence
- The study evaluated methods on two datasets: one for prostate cancer with 322 samples and another for ovarian cancer with 91 samples.
- Consensus ranking improved classification accuracy compared to individual feature ranking methods.
- The proposed methods can be applied to other large-scale experiments beyond mass spectrometry.
Takeaway
This study shows that using multiple methods to choose important features from mass spectrometry data can help doctors better identify cancer.
Methodology
The study applied several feature ranking procedures and computed a consensus list of features based on their outcomes, validated on two proteomic datasets.
Limitations
The study primarily focused on mass spectrometry data and may not generalize to other types of data.
Participant Demographics
322 samples from prostate cancer patients and healthy donors, and 91 samples from ovarian cancer patients and controls.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website