-
Notifications
You must be signed in to change notification settings - Fork 1
panelMSI
The MismatchResolveDx panel specifically targets mono-nucleotide microsatellite regions known to experience frequent deletions in mismatch repair deficient tumours. We have therefore developed a method to examine these regions in order to classify a sample as MSS or MSI.
-
provide a bed file of your targeted regions; this will identify which regions cover mono-nucleotide repeats (minimum length of 13 nucleotides) and annotates these with i) the repeated nucleotide, ii) length of the microsatellite, iii) start and end position of the microsatellite in the reference genome
-
provide the above detected regions and a BAM file; for each microsatellite identified above, this will i) extract all reads from the BAM file that overlap the region, ii) determine the length of the microsatellite in each read, and iii) return the read counts for each observed length
-
provide the readcounts generated in #2 for all 'normal' samples; for each region, find the median count of reads containing i) deletions, ii) insertions or iii) the expected (reference) allele
-
provide the readcounts generated in #2 for all 'tumour' samples and the panel of normals from #3; for each region, compare the proportion of reads with a deletion in each tumour sample to the panel of normals
The final output table from runPanelMSI.R provides various metrics for each region, including the proportion of reads with a deletion in the tumour BAM, the proportion of reads with a deletion in the panel of normals and a p-value from a one-sided proportions test (is the proportion of reads with a deletion higher in the tumour than in the panel of normals). collect_panelMSI_output.R will identify the fraction of regions with p < 0.01 and applies a pre-determined threshold to classify the tumour sample as MSS or MSI.
The threshold used for classification was determined by running the above on 53 'normal' samples (a threshold of 0.25 was sufficient to classify all normal samples MSS).