Different results for same SMILES in different sdf files, using PAINS.sieve -- why?

[Kirk Ehmsen]

Hi,

Can someone help me with 2 questions about using filter-it at the command line?  

1.  When I use filter-it to flag SMILES with PAINS fragments (file type .sdf), SMILES flagged as containing PAINS fragments in one file don't always come up as flagged when they're processed in a second file.

Details:

I generated the .sdf files from .smi files, using obabel:

SDF file 1

obabel /Users/kirkehmsen/Downloads/filter-it-1.0.2/SMILES1.smi -O /Users/kirkehmsen/Downloads/filter-it-1.0.2/SMILES1.sdf --gen3d

SDF file 2

obabel /Users/kirkehmsen/Downloads/filter-it-1.0.2/SMILES2.smi -O /Users/kirkehmsen/Downloads/filter-it-1.0.2/SMILES2.sdf --gen2d

If I run filter-it on SMILES1 file, I get two SMILES flagged as pyrroles:

/usr/local/bin/filter-it --input='/Users/kirkehmsen/Downloads/filter-it-1.0.2/SMILES1.sdf' --filter='PAINS.sieve'

OC(=O)c1ccc(cc1O)n1c(C)ccc1C >> failed FRAGMENT "regId=pyrrole_A(118)" filter criterion (2)

OC(=O)CSc1ccc(cc1)n1c(C)ccc1C >> failed FRAGMENT "regId=pyrrole_A(118)" filter criterion (2)

However if I run filter-it on SMILES2 file, these two same SMILES are not flagged as pyrroles:

/usr/local/bin/filter-it --input='/Users/kirkehmsen/Downloads/filter-it-1.0.2/SMILES2.sdf' --filter='PAINS.sieve'

OC(=O)c1ccc(cc1O)n1c(C)ccc1C >> passed

OC(=O)CSc1ccc(cc1)n1c(C)ccc1C >> passed

** Note that the two .sdf files were processed with different 2d vs. 3d functions in obabel.  When I tried to use 3d functions for SMILES2 file, I encountered an error reading: WARNING: damped steplength 0.00868418 to 0.00597873

It looks like the different 2d vs. 3d processing for the sdf files makes a (perhaps 'the') difference here -- but why?  And what is a likely reason to encounter a 'damped steplength' error that makes the 3d function fail for one of my two sdf files?

2.  Also, a second question -- if I ask the file to be run in tabular form, neither SMILES1 or SMILES2 files has these SMILES flagged:

/usr/local/bin/filter-it --input='/Users/kirkehmsen/Downloads/filter-it-1.0.2/SMILES1.sdf' --filter='PAINS.sieve' --tab

OC(=O)c1ccc(cc1O)n1c(C)ccc1C >> passed

OC(=O)CSc1ccc(cc1)n1c(C)ccc1C >> passed

Why doesn't the PAINS filter work now when --tab is used?

Thank you for any help with these questions.

Thank you,

Kirk

UCSF