I'm using break_blocks to extract regions of a gVCF file:

zcat NA19238.final_combined.g.vcf.gz | break_blocks --region-file file.bed --exclude-off-target --ref /illumina/runs/con/g1k_v37/human_g1k_v37.fasta > input.g.vcf

which produces a file with about 6 of the bed files's 43 sites, which isn't very good.

the tab-delimited bed file looks like:

10  96405502    96405502    label1
10  96541616    96541616    label2
10  96540410    96540410    label3

I get similar problems when the bed file looks like this:

10  96405502    96405503    label1
10  96541616    96541617    label2
10  96540410    96540411    label3

Almost all of the sites are missing, and yet because of the GVCF format, nearly all should be present, why are so many spots missing, then?

Also, when I convert gVCF to VCF by GATK's tool

gatk --java-options "-Xmx4g" GenotypeGVCFs -R human_g1k_v37.fasta -Vinput.g.vcf.gz -O output.final_combned.vcf.gz

I get no lines at all, why is this happening?