Reviewing fits

1. Loading facets runs

There two ways to load fits into facets-preview. The first is "repo" mode where samples are input through a manifest file. This file contains at least three columns (and up to four). See manifest file requirements. Second option is "sample" mode which simply accepts one or more sample-level facets output directories.

2. Samples manifest

The samples manifest page gives a high-level overview of the QC status of the default fit as well as that of the reviewed fit (if it exists).

3. Review fits

3.1 Submit refits

• Refits require the watcher to be running. See instructions for setting up watcher. If there is a "red cross" it means the watcher is not running.
• Pre-filled options for all the refit parameters are from the default fit (if there is no default fit, then options used to generate any of the fits are used). Note: the compatible facets versions can be configured in the config file.
• Note: It takes a few mins to generate the refits. Simply, go to the "Samples Manifest" page and click on the sample again to reload with newly generated fits.

3.2 QC Summary

The QC for each fit is generated by an facets_qc.R script in the facets-preview package (https://github.com/taylor-lab/facets-preview/tree/master/facets_qc/) that is independently version tracked. The purpose is to enable faster evolution of the QC criteria. For more information on QC criteria, see FACETS QC criteria section.

The "QC metrics" are simply the aggregation of facetsSuite::check_fit metrics as well as the facets_qc flags. All columns are also accessible in the sample-level facets_qc.txt file.

3.3 Close Up

3.4 Segments

Unedited segment output of the facets algoirthm

3.5 Segments (editable)

Same data as in 3.4 but allows editing the cncf.txt file. Useful for correcting TCN/LCN values for certain segments.

3.6 Review notes

facets-preview enables annotating samples at multiple levels.

• "Acceptable fit" and "Best fit" differ only by subjective assessment of analyst in degree of confidence in the fit. Currently, no distinction is made between the two in calculating genomic annotations. Therefore, it is up to the analyst to take this distinction in consideration in the downstream analysis.
• "Use purity run for CCF annotation". In some cases, a hisens run is undesirable (eg: too high fragmentation, etc) and a purity run is preferable for CCF calculations.

• "Use purity". In cases where purity is NA/0.3, this allows analyst to enter a purity value estimated from mutations (for example, MSI tumors are flat and therefore yield a purity of NA).

• Note: In case of multiple manual reviews, the one that is most recent is used for genomic annotations.