MTAP add-on workflow resources

A published application note describes a proof-of-concept study using a spike-in probe set with the TSO 500 product and the DRAGEN Enrichment BSSH App 4.4.6 to enable MTAP gene CNV calling. This page describes the data and resources that were used in the app note. The resources and output on this page were generated using the reference genome Homo sapiens [NCBI] hs37d5 with chr prefix v5.

Probe Information

Probes for the MTAP spike in were designed to cover exonic and intronic locations that are unique in the genome and did not interfere with OPD2 (TSO 500) or OPD3 (HRD) probe sets. The probes are optimized for CNV calling and do not cover all exonic regions.

In addition to the MTAP targets, the MTAP spike-in design includes intronic probes for ALK and ESR1 to serve as internal controls. Because ALK and ESR1 exons are targeted in OPD2, the spike-in probes targeting these genes' introns enable the user to check if the spike in concentration was comparable to the OPD2 probe concentration (see details in the QC section below).

The probe pool has 196 unique probes: 146 MTAP, 25 ALK, 25 ESR1.

Resource files

Use the below resource files to run the DRAGEN Enrichment App in BSSH or using command line. Refer to Analyzing with DRAGEN Enrichment App for instructions. Note that these resources were produced using Homo sapiens [NCBI] hs37d5 with chr prefix v5 and are only intended for running DRAGEN Enrichment using that reference genome.

Targeted Regions bed file: The manifest inclusive of TSO 500 and the MTAP spike-in targets. Provide in Custom Target BED File field and in Additional Resource Files under Additional Arguments in the BSSH App.

TSO500_PLUS_MTAP_targets_noheader.bed

489KB

Open

Target Segmentation bed: The bed file with the gene boundaries for MTAP for segment mean calling. Provide in Additional Resource Files under Additional Arguments in the BSSH App.

MTAP_seg_only.bed

27B

Open

Panel of Normals: Counts from 22 normal run with TSO 500 + MTAP probes. Provide in Additional Resource Files under Additional Arguments in the BSSH App.

MTAP_PON_525_20260428.combined.counts.txt.gz

560KB

Open

QC Bed File: Segmentation bed of ALK and ESR1 exon and intronic boundaries to be used for quality control. Provide in Additional Resource Files under Additional Arguments in the BSSH App.

QC_seg_housekeeping_MTAP.bed

3KB

Open

Additional Command-line Arguments: In this BSSH field (under Additional Arguments), provide the following arguments:

--enable-cnv true --cnv-filter-copy-ratio 0.2 --cnv-enable-filter-copy-ratio true --cnv-filter-qual 25 --cnv-enable-gcbias-correction false --cnv-enable-ref-calls true --cnv-enable-split-intervals false --cnv-normalize-rows-by-interval-size true --cnv-linear-regression-noise-removal true --cnv-normalize-counts-by-good-median true --cnv-count-mapq-zero false --cnv-num-singular-values 5 --cnv-enable-tso500-arm-level-detection true --cnv-segmentation-mode bed --cnv-enable-tso500-gender-call true --cnv-target-bed TSO500_PLUS_MTAP_targets_noheader.bed --cnv-segmentation-bed MTAP_seg_only.bed --cnv-combined-counts MTAP_PON_525_20260428.combined.counts.txt.gz --cnv-filter-duplicate-alignments false --enable-duplicate-marking false --qc-coverage-region-1 QC_seg_housekeeping_MTAP.bed --qc-coverage-tag-1 QC --qc-coverage-reports-1 cov_report full_res

Running DRAGEN Enrichment for MTAP analysis on Local DRAGEN Server

If you are planning to run on a local DRAGEN server, use the below command and resources. Note that you may need to contact customer support for assistance with the specific DRAGEN license needs for running DRAGEN Enrichment.

Command:

dragen 
--vc-use-somatic-hotspots true \
--vc-frd-max-effective-depth 40 \
--cnv-enable-ref-calls true \
--output-format BAM \
--cnv-enable-gcbias-correction false \
--enable-variant-caller true \
--umi-enable true \
--umi-library-type random-duplex \
--qc-coverage-region-padding-2 150 \
--vc-bqd-max-effective-depth 50 \
--enable-map-align-output true \
--umi-min-supporting-reads 2 \
--enable-vcf-compression true \
--vc-target-bed TSO500_PLUS_MTAP_targets_noheader.bed \
--enable-metrics-json true \
--vc-systematic-noise NONE\
--vc-target-bed-padding 0 \
--enable-map-align true \
--vc-enable-vcf-output true \
--vc-enable-umi-solid true \
--vc-emit-ref-confidence GVCF \
--umi-metrics-interval-file TSO500_PLUS_MTAP_targets_noheader.bed \
--vc-skip-germline-tagging true \
--enable-duplicate-marking false \
--logging-to-output-dir true \
--output-file-prefix $SAMPLEID \
--output-directory $SAMPLEID \
--tumor-fastq-list tumor-fastq-list.csv \
--tumor-fastq-list-sample-id $SAMPLEID \
--sample-sex none \
--generate-ploidy-vcf false \
--ref-dir \/scratch\/hs37d5_chr-cnv\/DRAGEN\/11 \
--qc-coverage-region-1 QC_seg_housekeeping_MTAP.bed \
--qc-coverage-filters-1 mapq<0 \
--qc-coverage-region-2 TSO500_PLUS_MTAP_targets_noheader.bed \
--qc-coverage-filters-2 mapq<0 \
--enable-cnv true \
--cnv-filter-copy-ratio 0.2 \
--cnv-enable-filter-copy-ratio true \
--cnv-filter-qual 25 \
--cnv-enable-split-intervals false \
--cnv-normalize-rows-by-interval-size true \
--cnv-linear-regression-noise-removal true \
--cnv-normalize-counts-by-good-median true \
--cnv-count-mapq-zero false \
--cnv-num-singular-values 5 \
--cnv-enable-tso500-arm-level-detection true \
--cnv-segmentation-mode bed \
--cnv-enable-tso500-gender-call true \
--cnv-target-bed TSO500_PLUS_MTAP_targets_noheader.bed \
--cnv-segmentation-bed MTAP_seg_only.bed \
--cnv-combined-counts MTAP_PON_525_20260428.combined.counts.txt.gz \
--cnv-filter-duplicate-alignments false \
--qc-coverage-tag-1 QC \
--qc-coverage-reports-1 cov_report full_res \
--qc-coverage-reports-2 full_res

Results from DRAGEN Enrichment App

The fold change (also known as segment mean or SM) can be found in the following location in the DRAGEN enrichment App output: $SAMPLEID/$SAMPLEID.cnv.vcf.gz. A single entry in the vcf is expected that will correspond to the MTAP segment mean. An example:

#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	10847
chr9	21802574	DRAGEN:REF:chr9:21802574-21867130	N	.	3	cnvQual	END=21867130;REFLEN=64557;SEGID=target-wes-chr9-21802573:21867130	GT:SM:BC:PE	./.:1.06262:100:0,3

While customers should interpret these results within the context of their own validation studies, our development work showed that combining fold change (SM) with tumor fraction estimates from the HRD probes provides the most accurate assessment of MTAP status (see Figure 8 of the Application Note). Accordingly, we do not rely on ALT DEL/DUP calls in the VCF.

Using the Housekeeping Genes included as QC

A known risk of this workflow is improper spike-in of the MTAP probes, leading to uneven coverage between TSO 500 and the MTAP target sequence and inaccurate fold change estimates. To reduce this risk, intronic sequence for two TSO 500 genes are targeted by the spike-in probes. If the spike-in was effective, we expect the exon:intron mean coverage ratio to be close to 1. In order to compute the spike in ratio, the $SAMPLEID/$SAMPLEID.QC_cov_report.bed from the DRAGEN Enrichment App output can be used.

Using the coverage report and the QC_seg_housekeeping_MTAP.bed resource, you can calculate the ratio for both genes. Because copy alterations are not likely to occur in both ALK and ESR1 in a single cancer type, at least one should be usable for QC even if a true biological CNV is observed in the other (to check for whole gene CNVs on these two genes you can look to the standard TSO output).

An example R script for calculating the ratio:

library(tidyverse)


#required 
resource
seg_housekeeping <- ("QC_seg_housekeeping_MTAP.bed")


#function
intron_exon_mean_ratio <- function(cov_path) {
                seg <- read_tsv(seg_housekeeping, col_names = c("chrom","start","end","seg_name"), show_col_types = FALSE)
  
                
                cov <- read_tsv(cov_path, show_col_types = FALSE, col_names = T) %>%
                
rename(chrom = `#chrom`)
  
                
                joined <- seg %>%
    
                inner_join(cov, by = c("chrom","start","end")) %>%
                select(c(chrom, start, end, seg_name, mean_cvg)) %>%
                mutate(region_type = if_else(str_detect(seg_name, regex("intron", ignore_case = TRUE)), "intron", "exon")) %>%
                mutate(gene = case_when(str_detect(seg_name, regex("^ALK",  ignore_case = TRUE)) ~ "ALK", str_detect(seg_name, regex("^ESR1", ignore_case = TRUE)) ~ "ESR1"))
  
                
                out <- joined %>%
                group_by(gene, region_type) %>%
    
                summarize(mean_cvg_region = mean(mean_cvg, na.rm = TRUE), .groups = "drop") %>%
    
                tidyr::pivot_wider(names_from = region_type, values_from = mean_cvg_region) %>%
    
                mutate(intron_exon_ratio = intron / exon) %>%
    
                arrange(gene)

  
                
                out}
                


#usage of function


intron_exon_mean_ratio("Sample.QC_cov_report.bed")

A spike-in that will reliably produce an accurate MTAP fold change will have ratios near one, such as Runs 1 and 2 below. If the ratio is much lower or higher than 1, as observed in Run 3 below, the MTAP fold change will not be accurate. In our experience, the ESR1 ratio appears to be more accurate and reproducible than the ALK ratio.