A machine learning approach to optimizing cell-free DNA sequencing panels: with an application to prostate cancer
This repository contains supporting material referenced in the publication with the above title. The manuscript can be found both on bioarxiv and in an upcoming publication.
- bioarxiv doi: https://doi.org/10.1101/2020.04.30.069658
- Upcoming publication: TBD
Key Files
-
comparison_panels/
Contains variant location compositions for the frequency, union-existing, and orchid panels. -
comparison_panels/orchid_panel.bed
A bed file containing variants used to generate the hybrid capture probes for the orchid generated panel. -
comparison_panels/supporting
Contains panel supporting reference materal and parsing scripts. -
notebooks/panel_generation.ipynb
A jupyter notebook used to generate the orchid panel, includes Figures 1A,1B,1C, and S1 -
notebooks/panel_evaluation.ipynb
A jupyter notebook used to assess the orchid panel, includes Figures 2A,2B,2C,2D, and S2 -
notebooks/reanalysis.ipynb
A jupyter notebook used to assess orchid panel in silico performance and to model one-vs-rest feature importance, includes Figures 1D and 3 -
notebooks/panel_performance.ipynb
A jupyter notebook to assess the number of patient variants captured by the panel, includes Figures 4A and 4B
Notes
Some folders referenced in the above notebooks are not present in this repository due to their large sizes and/or patient confidentiality.
-
TN/
Contains confidential patient tumor-normal variants -
cfDNA/
Contains confidential patient cfDNA detected variants -
SVMv4_Enrichment_20171107
Contains the SVM model used to generate the orchid panel, available upon request