Indeed - adding 2d coordinates fixed comparison based on InChI, however
Morgan fingerprints still show some differences though I need to check
whether it's for the same 26 examples.
My whole set is larger and it will take a bit of digging to find out.
I guess I'm still missing something...
Thanks!
Igor
On Mon, Jan 27, 2014 at 11:42 PM, Greg Landrum <[email protected]>wrote:
> hmm, I can't reproduce that:
>
> In [16]: inls = [x.strip().split() for x in
> file('/Users/landrgr1/Downloads/t.tab').readlines()]
>
> In [17]: inls.pop(0)
> Out[17]: ['InChI', 'Original_InChI']
>
> In [18]: for inch,o_inch in inls:
> ....: m = Chem.MolFromInchi(o_inch)
> ....: n_inch = Chem.MolToInchi(m)
> ....: if n_inch!=o_inch:
> ....: print o_inch
> ....: print n_inch
> ....: print Chem.MolToSmiles(m,True)
> ....:
> [05:37:40] WARNING: Proton(s) added/removed
> [05:37:41] WARNING: Proton(s) added/removed
> [05:37:41] WARNING: Proton(s) added/removed
> [05:37:41] WARNING: Charges were rearranged
>
> In [19]: from rdkit.Chem import AllChem
>
> In [20]: for inch,o_inch in inls:
> ....: m = Chem.MolFromInchi(o_inch)
> ....: AllChem.Compute2DCoords(m)
> ....: mb = Chem.MolToMolBlock(m)
> ....: nm = Chem.MolFromMolBlock(mb)
> ....: n_inch = Chem.MolToInchi(nm)
> ....: if n_inch!=o_inch:
> ....: print o_inch
> ....: print n_inch
> ....: print Chem.MolToSmiles(m,True)
> ....: print Chem.MolToSmiles(nm,True)
> ....:
> [05:39:44] WARNING: Proton(s) added/removed
> [05:39:44] WARNING: Proton(s) added/removed
> [05:39:44] WARNING: Proton(s) added/removed
> [05:39:44] WARNING: Charges were rearranged
>
> In [21]:
>
> Could it be that you forgot to add coordinates before you generated the SD
> file?
>
> -greg
>
>
>
> On Tue, Jan 28, 2014 at 5:22 AM, Igor Filippov
> <[email protected]>wrote:
>
>> Here are some examples - "original InChI" were created from the original
>> SD file, then a new SD file was created from those and new InChI
>> calculated, called here "InChI".
>> It's a tab-separated table.
>> Thanks for taking time to look at this!
>>
>> Igor
>>
>>
>> On Mon, Jan 27, 2014 at 10:55 PM, Greg Landrum <[email protected]>wrote:
>>
>>> On Tue, Jan 28, 2014 at 2:49 AM, Igor Filippov <
>>> [email protected]> wrote:
>>>
>>>> I noticed that if I convert mol to inchi and then back to mol in quite
>>>> a few cases
>>>> the stereochemistry information gets lost. Is it something that is
>>>> handled completely by InChI library
>>>> or is RDKit not reading the mols produced from InChI correctly?
>>>>
>>>
>>> It could be problems in the RDKit conversion to or from InChI. The
>>> easiest way to check where the problem is coming from is to see if the
>>> InChI itself has the correct stereochemistry flags. If not, it's the
>>> RDKit->InChI process, otherwise it's InChI->RDKit.
>>>
>>> Feel free to send along some example molecules if you want me to take a
>>> look at them.
>>>
>>> Best,
>>> -greg
>>>
>>
>>
>
------------------------------------------------------------------------------
WatchGuard Dimension instantly turns raw network data into actionable
security intelligence. It gives you real-time visual feedback on key
security issues and trends. Skip the complicated setup - simply import
a virtual appliance and go from zero to informed in seconds.
http://pubads.g.doubleclick.net/gampad/clk?id=123612991&iu=/4140/ostg.clktrk
_______________________________________________
Rdkit-discuss mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
