In my original email, in an effort to simplify the comparison I introduced some rounding errors. The correct output is: MR In OB, not in RDKit: set([-0.57540000000000002]) In RDKit, not in OB: set([5.7539999999999996]) LOGP In OB, not in RDKit: set([0.084500000000000006, 0.081299999999999997, 0.14410000 000000001, 0.083900000000000002, -1.95, 0.01508]) In RDKit, not in OB: set([-0.19500000000000001, 0.08387, 0.081290000000000001, 0 .084519999999999998, 0.1411, 0.010580000000000001])
There's an error in the MR data in OB (Me1) - I've corrected this in SVN. There are three cases in OB where the LogP data has been shortened to 3 decimal places (instead of the 4 that are in the paper), leading to the disagreement with RDKit, but I don't think these are worth changing. Noel On 24/04/2008, Greg Landrum <[email protected]> wrote: > Noel, > > Thanks for doing this; it's a nice way to catch typos. > > > On Thu, Apr 24, 2008 at 9:03 AM, Noel O'Boyle <[email protected]> wrote: > > MR > > In OB, not in RDKit: set([-0.57540000000000002]) > > In RDKit, not in OB: set([5.7538999999999998]) > > > This is Me1 and is correct in the RDKit. > > > > LOGP > > In OB, not in RDKit: set([0.083900000000000002, -1.95, > 0.081299999999999997, 0.0 > > 14999999999999999, 0.14410000000000001]) > > In RDKit, not in OB: set([0.081199999999999994, 0.1411, > 0.010500000000000001, 0. > > 083799999999999999, -0.19500000000000001]) > > > > The 0.1411 vs. 0.1441 error (LogP) in RDKit was identified yesterday > > or so by Greg. > > > I don't find 0.081199... in my parameter file. I do find 0.08129, > which is correct for CS > 0.1411 is the C1 typo (as you mention above) > 0.0105 I don't find, though I do see 0.010508 (N9), which should be 0.01508 > 0.08379... I don't find, it's probably 0.08387 (N5), which is correct. > -0.195 is N10, and should be -1.95(!!!) . this is the first of these > that will make a substantial difference. > > I will be fixing these in the RDKit for the next release. > > > -greg >
