search

arq5x / lumpy-sv

lumpy: a general probabilistic framework for structural variant discovery
MIT License
314 stars 118 forks source link

lumpy returns more than expected variants #382

Open anoronh4 opened 2 years ago

anoronh4 commented 2 years ago

i have run lumpy express as follows: 

singularity exec \
    -e --no-home \
    -B /path/to/bams/ \
    ../lumpy_v0.2.13-2d611fa.sif \
    lumpyexpress \
      -B ${bamTumor},${bamNormal} \
      -S ${idTumor}.splitters.bam,${idNormal}.splitters.bam \
          -D ${idTumor}.discordants.bam,${idNormal}.discordants.bam \
      -o ${idTumor}__${idNormal}.vcf

I am finding that in our test, lumpy generates a lot more passing variants compared to other callers, for one sample we have less than 200 variants in each of 3 other callers, and lumpy produces >10000. just wondering if there was an acceptable way to filter the calls, particularly for tumor/normal pairs where we are trying to find variants specific to the tumor.