random question im looking at the description of the pipeline. so you prepare a sample sheet with svs,cnvs,snvs. you need to run it 3 times with a truth svs, truth cnvs, truth snvs? is this a correction interpretation? and I would just swap params.vartype for each test? does that mean it ignores the "non-relevant" variant types in the sample sheet?

basically, you need to prepare your runs per variant type (either for snv, small, indel, sv, cnv). For SV benchmarking, you can prepare your samplesheet accordingly with tests sv calls, choose "variant_type=''structural'" provide your truth in truth_vcf and make your run. https://github.com/nf-core/variantbenchmarking/blob/master/conf/tests/somatic_sv.config. plenty of different runs here https://github.com/nf-core/variantbenchmarking/tree/master/conf/tests explained https://github.com/nf-core/variantbenchmarking/blob/master/docs/testcases.md

cool thanks for the reply