runABS

A collection of R scripts and functions to run ABSOLUTE on pipeline mutation summary outputs

Generating a standard package summary

Metrics file

1. Check that none of the tumor samples have sequencing statistics below 90

geneCN

1. Go through each and every sample and check whether amplifications or homozygous deletions were called correctly and ensure their prsense in the sample geneCN.txt and visible in the copy number corresponding plot
   • our methods allow for manual curation, thus, inspect and confirm all regions marked as homozygous deletions. Some of these may need to be changed to a loss if the signal lacks strength
2. FACETS can miss calls depending on how the genome is segmented. These parameters change the sensitivity and size of CNV detected. If there are regions that should be considered having a CNV please discuss it with the group as to ascertain what parameters should be modified to improve our CNV sensitivity.
3. Regions as true homozygous deletions or amplifications need to be screened for cancer genes from our three lists

Mutation Summary

1. From the summary excel file, save the MUTATION_SUMMARY sheet to a .csv file, we suggest muts.csv
2. Move the FACETS *.cncf.txt files from each sample to the same directory as muts.csv
3. Run the script Make_ABS_muts_Input.R to generate per sample input files for ABSOLUTE
4. Run Make_SegFromFacets.R to create the copy number input files for ABSOLUTE
5. Run RunAbsolute.R
   1. If you run this script in the same R session as the previous two, sampleList object is preserved and will run the ABSOLUTE on a loop
   2. Run Step 1 and Step 2 of RunAbsolute.R
   3. An Output/ directory containing absolute_results/ will be created
   4. Select ABSOLUTE solutions for each case.
   • In Output/absolute_results/ open the *.PP-calls_tab.txt file.
   • Create a new first column labeled "solution"
   • Open the sample .pdf from Output/absolute_results/ .
     • Best solution should have an SSNV multiploidy peak between 1.0 or 2.0 and CCF plot should have a green peak around 1.0
   • If you are undecided between two solutions, pick the lower number solution and consult with a colleague 5.Once solutions are set run Step 3 in RunAbsolute.R
6. Run CombineAbsMaf.R
   1. Change the MAF directory to where the results from ABSOLUTE are present. This is typically Output/absolute_results/Output/reviewed/SEG_MAF
   2. Asign to results.directory the location of the muts.csv file
   3. This is all annotated in the code. If the hotspots are already annotated you do not need to redo this step. Currently, REDO the LOH
7. QC pipeline LOH and hotspot annotation on the csv file
   1. Compare first iteration HOTSPOT loci calls by the pipeline to the hostpot loci calls fromm these R scripts. These should be identical, however, discrepancies occur when the ANN....EFFECT is a splice site mutation, intronic variants, and the gene has known hotspot loci.
   2. Compare LOH pipeline calls by the pipeline to loh calls from these R scripts.
   • When mutations fall outside segments these are marekd as errors and should be curated by hand from the cncf/ segment files.
     • use the segment closest to the mutation position, where if lesser copy number (lcn) = 0 is LOH and lnc ≥ 1 is NO-LOH.
   3. Make sure there are no missing values from the ABSOLUTE calculations or any other required fields
8. Run Mutation_Figure_Patient.R
   1. Comments and suggestions throughout the code
   2. Will need to make cosmetic chages to the figures
   • In the CCF heatmap, surround mutations with yellow circles with a yellow box indicating these are clonal. Those with LOH need to have a white slash through them
   • Make all samples bold, mutations italicized and everything should be in Arial Font
9. Prepare Excel Spreadsheet summary
   1. Remove FACETS CCF and Clonal Status columns as ABSOLUTE CCF and Clonal Status tend to be more accurate. However, compare these to see how much concordance we have between the calls.
10. Run make mutsig_report