Re: [Rdkit-discuss] SMILES string from SureChEMBL iPython Notebook Tutorial

2016-07-05 Thread DmitriR
... sounds like a good explanation. If this is indeed the case, it would be reasonable to force flush to get guaranteed non-random behavior. Thanks! Best, Dmitri > On Jul 5, 2016, at 7:39 AM, Brian Kelley wrote: > > After some digging, it looks like the underlying C++

Re: [Rdkit-discuss] Incorrect SMARTS from FindMCS

2016-07-05 Thread Greg Landrum
On Tue, Jul 5, 2016 at 1:38 PM, Peter S. Shenkin wrote: > But we have learned that RDKit can do the imparsable > > Oh... Peter... you should be ashamed. :-) -- Attend Shape: An AT Tech Expo July

Re: [Rdkit-discuss] Incorrect SMARTS from FindMCS

2016-07-05 Thread Peter S. Shenkin
But we have learned that RDKit can do the imparsable -P. On Tue, Jul 5, 2016 at 6:14 AM, Marta Stępniewska-Dziubińska < mart...@ibb.waw.pl> wrote: > Hi Greg, > > To be honest, I expected it to fail :) > > I want to find structures with a defined topology, but allow for > different bond

Re: [Rdkit-discuss] SMILES string from SureChEMBL iPython Notebook Tutorial

2016-07-05 Thread Brian Kelley
After some digging, it looks like the underlying C++ streams aren't flushing. This means that python might not actually have all the information when you print them out. We may have to enable a "flush" function for these streams for better error reporting on the python side, I'll need to

Re: [Rdkit-discuss] Incorrect SMARTS from FindMCS

2016-07-05 Thread Marta Stępniewska-Dziubińska
Hi Greg, To be honest, I expected it to fail :) I want to find structures with a defined topology, but allow for different bond types. My first idea was to use SMARTS with ~ bonds, and it works fine with GetSubstructMatch. I don't really need MCS, I just used it to check different patterns. But

Re: [Rdkit-discuss] Incorrect SMARTS from FindMCS

2016-07-05 Thread Greg Landrum
Hi Maria, Output 11 is certainly wrong (the SMARTS generated is incorrect). I suspect this is happening because the molecule "coa" (input 3 in your notebook) is constructed from SMARTS and includes ~ bonds (which match any bond type). I almost think that the MCS code should be generated to fail

[Rdkit-discuss] Incorrect SMARTS from FindMCS

2016-07-05 Thread Marta Stepniewska-Dziubinska
Hi all, I got a very surprising error: RDKit generated a SMARTS string that it cannot parse. Here's my code: https://goo.gl/ETvjjT In order to explain the code: I create a structural pattern with SMARTS. It's a strange pattern, I admit, but I was able to find it in my molecule. I then try to