Potential bug in LogTool::prob2pl

[dwuab]

This issue has been raised in hyunminkang/vt-topmed#1, but it seems that vt-topmed is an archive repo.
I found a problem in prob2pl from log_tool.cpp:

uint32_t LogTool::prob2pl(double x)
{
    if (x>1 || x<0)
    {
        std::cerr << "[e] x is not a probability\n";
        exit(1);
    }

    return (uint32_t) (round(-10*std::log10(x)));
}

If x is 0.0, round(-10*std::log10(x)) is inf as expected, but the subsequent conversion to uint32_t make the returned value 0, which is clearly wrong.
This could affect genotype calling. In my case (genotype calling of HG002,HG003,HG004 WES samples from Genome In A Bottle), this results in homozygous ref genotype erroneously called as homozygous alt. To cite an example, see

20      126155  .       GCAAA   G       255     .       AVGDP=363;AC=6;AN=6;AF=1;GC=0,0,3;GN=3;HWEAF=1;HWDGF=1.09424e-09,2.70797e-24,1;IBC=1;HWE_SLP=0;ABE=0.562105;ABZ=2.75288;BQZ=0;CYZ=0.586203;STZ=0.000924303;NMZ=0.254606;IOR=0.998043;NM1=0.0372093;NM0=0.0317044        GT:GQ:AD:DP:PL  1/1:255:255,0:283:0,255,0       1/1:105:255,215:509:255,0,0     1/1:255:255,0:297:0,255,0

where you can see the HG002's PL being 0,255,0, which looks impossible.