Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Andrew Dalke]

And I have uploaded a source tar.gz and a binary wheel to PyPI.

That means you can do "pip install mmpdb" to install this most recent version.

Andrew
da...@dalkescientific.com



> On May 9, 2018, at 18:04, Kramer, Christian  
> wrote:
> 
> Dear all,
> 
> Andrew's fix for the wildcard atom representation in RDKit 2018.3 has just 
> been incorporated into the main mmpdb branch. If you download the latest 
> version of mmpdb, it should now work with with the current version and also 
> with previous RDKit versions.
> 
> Note that you have to rebuild your mmp database if you switch from a pre 2018 
> version of RDKit to a 2018+ version.
> 
> Bests,
> Christian
> 
> Dr. Christian Kramer
> 



--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Kramer, Christian]

Dear all,

Andrew's fix for the wildcard atom representation in RDKit 2018.3 has just
been incorporated into the main mmpdb branch. If you download the latest
version of mmpdb, it should now work with with the current version and also
with previous RDKit versions.

Note that you have to rebuild your mmp database if you switch from a pre
2018 version of RDKit to a 2018+ version.

Bests,
Christian

*Dr. Christian Kramer*

Computer-Aided Drug Design (CADD) Senior Scientist


F. Hoffmann-La Roche Ltd

Pharma Research and Early Development
Bldg. 092/8.56 C

CH-4070 Basel


Phone +41 61 682 2471

mailto: christian.kra...@roche.com


*Confidentiality Note: *This message is intended only for the use of the
named recipient(s) and may contain confidential and/or proprietary
information. If you are not the intended recipient, please contact the
sender and delete this message. Any unauthorized use of the information
contained in this message is prohibited.

On Tue, May 8, 2018 at 4:04 PM, Andrew Dalke 
wrote:

> Dear Marco,
>
> > On May 7, 2018, at 23:59, Marco Stenta  wrote:
> > I had some time to set an environment for it and test it: it works fine,
> as far as my tests go. I will switch to this version and to the latest
> RDKIT now.
>
> Thanks for the feedback. Someone else sent me a private email also saying
> it worked. I'll put together the final changes for a 2.1 release, and see
> about making it accessible from PyPI so "pip install mmpdb" will work.
>
> > some questions:
> > Is there any plan to:
> > include MCS as a fragmentation method?
> > extend to matched series?
> > include "fuzzy" environment definitions based on pharmacophores (as BI
> people did)?
>
> I know of no plans for that.
>
> As a consultant, my answer is more along the lines of "are you willing to
> pay for it?" :)
>
> It won't be cheap.
>
> Speaking of which, many thanks to Roche because they funded this work,
> just like that previously funded me to develop the MCS code that is now in
> RDKit.
>
>
> > I am currently using the database file to explore the rules mainly via
> visual inspection through spotfire by means of a series of joins to
> generate suitable tables: would anybody be interested in this (also helping
> improving it)?
>
> I hope you find others and are able to get this out there.
>
> One of the reasons for developing what eventually became mmpdb 2.0 is to
> make this sort of viewer possible.
>
> That is, the unreleased mmpdb 1.0 didn't have canonical attachment point
> assignment, resulting in up to 6 different ways to represent a 3-cut
> fragment. This made it technically challenging to the simple sorts of
> analyses that mmpdb 2.0 does now, and impossible to visualize meaningfully.
>
> Cheers,
>
> Andrew
> da...@dalkescientific.com
>
>
>
> 
> --
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> ___
> Rdkit-discuss mailing list
> Rdkit-discuss@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
>
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Andrew Dalke]

Dear Marco,

> On May 7, 2018, at 23:59, Marco Stenta  wrote:
> I had some time to set an environment for it and test it: it works fine, as 
> far as my tests go. I will switch to this version and to the latest RDKIT now.

Thanks for the feedback. Someone else sent me a private email also saying it 
worked. I'll put together the final changes for a 2.1 release, and see about 
making it accessible from PyPI so "pip install mmpdb" will work.

> some questions:
> Is there any plan to:
> include MCS as a fragmentation method?
> extend to matched series?
> include "fuzzy" environment definitions based on pharmacophores (as BI people 
> did)?

I know of no plans for that.

As a consultant, my answer is more along the lines of "are you willing to pay 
for it?" :)

It won't be cheap.

Speaking of which, many thanks to Roche because they funded this work, just 
like that previously funded me to develop the MCS code that is now in RDKit.


> I am currently using the database file to explore the rules mainly via visual 
> inspection through spotfire by means of a series of joins to generate 
> suitable tables: would anybody be interested in this (also helping improving 
> it)?

I hope you find others and are able to get this out there.

One of the reasons for developing what eventually became mmpdb 2.0 is to make 
this sort of viewer possible.

That is, the unreleased mmpdb 1.0 didn't have canonical attachment point 
assignment, resulting in up to 6 different ways to represent a 3-cut fragment. 
This made it technically challenging to the simple sorts of analyses that mmpdb 
2.0 does now, and impossible to visualize meaningfully.

Cheers,

Andrew
da...@dalkescientific.com



--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Marco Stenta]

Dear Andrew,
I had some time to set an environment for it and test it: it works fine, as
far as my tests go. I will switch to this version and to the latest RDKIT
now.

Thanks a lot for your help.

some questions:
Is there any plan to:
include MCS as a fragmentation method?
extend to matched series?
include "fuzzy" environment definitions based on pharmacophores (as BI
people did)?

I am currently using the database file to explore the rules mainly via
visual inspection through spotfire by means of a series of joins to
generate suitable tables: would anybody be interested in this (also helping
improving it)?

Cheers,
marco

2018-05-07 14:50 GMT+02:00 Taka Seri :

> Dear Andrew,
>
> I installed following version of mmpdb with new version of rdkit.
> And it worked fine!
> I appreciate your work.
> Kind regards,
>
> Takayuki
>
> On Apr 27, 2018, at 00:20, Andrew Dalke  wrote:
> > Please try out:
> >  http://dalkescientific.com/mmpdb-2.1b1.tar.gz
> >
> > or my fork at:
> >  https://github.com/adalke/mmpdb
> >
> > and let me know of any problems.
>
> Has anyone downloaded and tested this code? I would like to push it to the 
> main
> repository and start the process of doing a new release.
>
> Cheers,
>
>
> 
> --
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> ___
> Rdkit-discuss mailing list
> Rdkit-discuss@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
>
>
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Taka Seri]

Dear Andrew,

I installed following version of mmpdb with new version of rdkit.
And it worked fine!
I appreciate your work.
Kind regards,

Takayuki

On Apr 27, 2018, at 00:20, Andrew Dalke  wrote:
> Please try out:
>  http://dalkescientific.com/mmpdb-2.1b1.tar.gz
>
> or my fork at:
>  https://github.com/adalke/mmpdb
>
> and let me know of any problems.

Has anyone downloaded and tested this code? I would like to push it to the main
repository and start the process of doing a new release.

Cheers,
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Andrew Dalke]

On Apr 27, 2018, at 00:20, Andrew Dalke  wrote:
> Please try out:
>  http://dalkescientific.com/mmpdb-2.1b1.tar.gz
> 
> or my fork at:
>  https://github.com/adalke/mmpdb
> 
> and let me know of any problems.

Has anyone downloaded and tested this code? I would like to push it to the main 
repository and start the process of doing a new release.

Cheers,

Andrew
da...@dalkescientific.com



--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Andrew Dalke]

On Apr 27, 2018, at 00:20, Andrew Dalke  wrote:
> It does not appear that the .fragment files also need to be redone, so 
> rebuilding the .mmpdb file is mostly a matter of re-running the index step.

I no longer think that is correct. While indexing will work, the resulting MMP 
database can no longer be used for making predictions.

When you upgrade to RDKit 2018, you'll need to rebuild your mmpdb data sets 
from scratch.

The same will be true any time the RDKit canonicalization algorithm changes.


Andrew
da...@dalkescientific.com



--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Andrew Dalke]

On Apr 26, 2018, at 12:38, Andrew Dalke  wrote:
> The automated mmpdb test suite isn't that good, so I still need to do some 
> manual testing. I won't be able to get to this until (hopefully) this evening.

I did that, and tracked down one more bug.

Please try out:
  http://dalkescientific.com/mmpdb-2.1b1.tar.gz

or my fork at:
  https://github.com/adalke/mmpdb

and let me know of any problems.

NOTE! Because of this change, anyone who switches from a pre-2018 RDKit to the 
most recent RDKit must rebuild the .mmpdb files.

It does not appear that the .fragment files also need to be redone, so 
rebuilding the .mmpdb file is mostly a matter of re-running the index step.




Andrew
da...@dalkescientific.com



--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Andrew Dalke]

On Apr 26, 2018, at 10:09, Marco Stenta  wrote:
> 
> Dear Colleagues,
> I just installed on conda env the new rdkit version
>  and wanted to try mmpdb but upon testing I got the error below
> reverting back to rdkit=2017.09.3.0 it works fine (I still get some errors 
> but it goes thrugh)
> 
> It is a bug, am I doing anything obviously wrong?

You did nothing wrong.

The latest version of RDKit changed the SMILES output for wildcard atoms from 
"*" to "[*]".

I missed this change in the release notes. It's further documented at 
  https://github.com/rdkit/rdkit/pull/1788

The mmpdb code does a lot of manipulation of the SMILES at the syntax level, 
and parts of the code expected that wildcard atoms would only be in the form 
"[*]".

I've put together a first go at a fix, which should work for both older and the 
newest versions of RDKit.

The automated mmpdb test suite isn't that good, so I still need to do some 
manual testing. I won't be able to get to this until (hopefully) this evening.

Until then, if you could try out

  http://dalkescientific.com/mmpdb-2.1b1.tar.gz
or my fork of the repository at
  https://github.com/adalke/mmpdb

and let me know if there are failures, that would help me with the testing.


Cheers,


Andrew
da...@dalkescientific.com



--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss

Re: [Rdkit-discuss] RDKIT 2018.3 and MMPDB problem

[Andrew Dalke]

On Apr 26, 2018, at 10:09, Marco Stenta  wrote:
> 
> Dear Colleagues,
> I just installed on conda env the new rdkit version
> and wanted to try mmpdb but upon testing I got the error below
> reverting back to rdkit=2017.09.3.0 it works fine (I still get some errors 
> but it goes thrugh)
> 
> It is a bug, am I doing anything obviously wrong?

You did nothing wrong.

The latest version of RDKit changed the SMILES output for wildcard atoms from 
"*" to "[*]".

I missed this change in the release notes. It's further documented at 
 https://github.com/rdkit/rdkit/pull/1788

The mmpdb code does a lot of manipulation of the SMILES at the syntax level, 
and parts of the code expected that wildcard atoms would only be in the form 
"[*]".

I've put together a first go at a fix, which should work for both older and the 
newest versions of RDKit.

The automated mmpdb test suite isn't that good, so I still need to do some 
manual testing. I won't be able to get to this until (hopefully) this evening.

Until then, if you could try out

 http://dalkescientific.com/mmpdb-2.1b1.tar.gz
or my fork of the repository at
 https://github.com/adalke/mmpdb

and let me know if there are failures, that would help me with the testing.


Cheers,


Andrew
da...@dalkescientific.com



--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss