Which vcf to use for tumor calls? (neuroblastoma)

I’m actually wondering more generally if there’s documentation about what the various files are and what purposes they are intended for.
But specifically, if I want to get all high quality tumor variant calls, what would be the best file to use? I’ve been looking at the data for one patient PT_YYGH8EMR. I was thinking that the somatic mutations found in the file 2114e40c-4db8-430e-a6c8-203f068037b2.consensus_somatic.PASS.vep.vcf.gz should be a subset of all the variants found in SL264139.hard-filtered.vcf.gz but it does not seem to be true. Out of 24 consensus somatic variants on Chr 21, only 2 are in the hard-filtered file.

So, what is the hard-filtered file useful for? Can I use it for my purpose of getting the tumor calls, or must I start from the bam or fastq? I’d really rather not spend a ton of money running these pipelines when it seems like this must have already been done.
Thank you!