Floating point exception in OpenBabel::OBGroupContrib::Predict for logP with PA0

[jlkjr]

• [ x] I believe this to be a bug with Open Babel
• This is a feature request

Environment Information

Open Babel version: 3.0.0
Operating system and version: Ubuntu 18.04

Expected Behavior

No floating point exception. I think there needs to be a check in vector3::normalize() for zero vectors since it is very hard to guarantee that no callers ever have a zero vector.

Actual Behavior

vector3::normalize() gets a zero vector.

Thread 3 "bfd-server-debu" received signal SIGFPE, Arithmetic exception.
[Switching to Thread 0x7ffff5fc2700 (LWP 9010)]
0x0000000000dc98c9 in OpenBabel::vector3::normalize() ()
(gdb) bt
#0 0x0000000000dc98c9 in OpenBabel::vector3::normalize() ()
#1 0x0000000000e0543f in OpenBabel::OBBuilder::GetNewBondVector(OpenBabel::OBAtom*, double) ()
#2 0x0000000000b4723b in OpenBabel::OBMol::AddNewHydrogens(OpenBabel::HydrogenType, bool, double)
()
#3 0x0000000000d1b06f in OpenBabel::OBGroupContrib::Predict(OpenBabel::OBBase*, std::__cxx11::basic_string<char, std::char_traits, std::allocator >*) ()

Steps to Reproduce

This happens with any PDB structure containing the PA0 ligand (e.g. 3E3Z), a phenyl ring with arsenic attached. Getting the logP property. Example code (the 'content' string contains a mol format string with the PA0 compound:

std::ios_base::openmode imode = std::ios_base::in|std::ios_base::binary;
std::ios_base::openmode omode = std::ios_base::out|std::ios_base::binary;
std::stringstream ssi(content, imode);
std::stringstream sso("", omode);
OpenBabel::OBConversion conv(&ssi, &sso);
OpenBabel::OBFormat* molformat = conv.FindFormat("mol");
conv.SetInFormat(molformat);
OpenBabel::OBMol mol;
molformat->ReadMolecule(&mol, &conv);
json js;
OpenBabel::OBDescriptor* pDesc = OpenBabel::OBDescriptor::FindType("logP");
if(pDesc) js["LogP"] = pDesc->Predict(&mol);

mol string for PA0.178:A:
PA0
JL BFD1119190613
Written by the BioLeap Fragment Displayer(TM)[BioLeap, Inc.]
12 12 0 0 0 0 0 0 0 0999 V2000
7.2100 -10.0520 2.6110 As 0 3 0 0 0 0 0 0 0 1 0 0
7.1560 -10.0390 1.2400 C 0 0 0 0 0 0 0 0 0 2 0 0
6.2490 -9.2320 0.5650 C 0 0 0 0 0 0 0 0 0 3 0 0
6.2170 -9.2220 -0.8210 C 0 0 0 0 0 0 0 0 0 4 0 0
7.0890 -10.0140 -1.5640 C 0 0 0 0 0 0 0 0 0 5 0 0
8.0000 -10.8330 -0.9030 C 0 0 0 0 0 0 0 0 0 6 0 0
8.0240 -10.8420 0.4890 C 0 0 0 0 0 0 0 0 0 7 0 0
8.6560 -11.4170 0.9490 H 0 0 0 0 0 0 0 0 0 8 0 0
8.6120 -11.3920 -1.4070 H 0 0 0 0 0 0 0 0 0 9 0 0
7.0620 -9.9950 -2.5330 H 0 0 0 0 0 0 0 0 0 10 0 0
5.5780 -8.6550 -1.2800 H 0 0 0 0 0 0 0 0 0 11 0 0
5.6350 -8.6730 1.0660 H 0 0 0 0 0 0 0 0 0 12 0 0
1 2 1 0 0 0 0
2 3 2 0 0 0 0
2 7 1 0 0 0 0
3 4 1 0 0 0 0
4 5 2 0 0 0 0
5 6 1 0 0 0 0
7 6 2 0 0 0 0
3 12 1 0 0 0 0
4 11 1 0 0 0 0
5 10 1 0 0 0 0
6 9 1 0 0 0 0
7 8 1 0 0 0 0
M END

[e-kwsm]

openbabel/src/math/vector3.cpp

Lines 152 to 161 in ac33dc4

	vector3& vector3 :: normalize ()
	{
	#ifdef OB_OLD_MATH_CHECKS
	if( CanBeNormalized() )
	(*this) /= length();
	#else
	(*this) /= length();
	#endif
	return(*this);
	}

Should always check if the norm is not zero.

[ghutchis]

Thanks for the detailed report.

Based on the backtrace, the bug is not in the vector code but in AddNewHydrogens / GetNewBondVector.