Tests on the CASF dataset

[kongmenghua]

Hello Zhang,

I conducted tests on the CASF dataset, but I noticed discrepancies between my test results and those reported in the literature. My Rp value is calculated to be 0.73, while the published results indicate a value of 0.824. I am considering the possibility that the manner in which I prepared the input files might be influencing the results. Therefore, I am seeking your guidance on the proper preparation of input files.

Currently, I prepare the protein structure file (.pdb) using prepwizard in Maestro. The molecule files (mols.sdf) are generated by first using openbabel to convert SMILES to a 2D SDF file, followed by preparation using epik in Maestro. This preparation involves adding hydrogen atoms and ensuring ionized states.Additionally, the ligand.sdf file is saved from the crystal structure.

Could you please advise if this approach might have any impact on the calculation results or if there are potential errors in my process?

[aquilazhang]

Hello! Thanks for your interests in PLANET. In fact, the core set has been mannually checked and processed by myself. Since some mol structures in the currently availble CASF-2016 on the PDBbind website has some problems (maybe bond type etc.) Here, I will upload a processed version of core set on the main page of this repository, on which I tested PLANET and acheive the published result. I has just re-conducted the scoring power test and the result is the same as reported (in the attachment). CASF_test.zip

[aquilazhang]

I have checked the code again, I found a mistake in the old version of PLANET_run.py when working with .sdf file (the chiral center feature are ignored. rdkit module can not read the chrial information from sdf generated by MAESTREO ). I have uploaded a revised version of PLANET_run.py. Or, you can transform the .sdf file from MAESTRO back to SMILES, this can work with the orginal code.

[kuku609]

Hello! Thanks for your interests in PLANET. In fact, the core set has been mannually checked and processed by myself. Since some mol structures in the currently availble CASF-2016 on the PDBbind website has some problems (maybe bond type etc.) Here, I will upload a processed version of core set on the main page of this repository, on which I tested PLANET and acheive the published result. I has just re-conducted the scoring power test and the result is the same as reported (in the attachment). CASF_test.zip

Hi, I would like to ask you how you dealt with small molecules in decoys_screening in CASF-2016 that could not be successfully read in by rdkit. I also encountered unreasonable bond issues when reading ligand molecules with rdkit, can you give me any advice, thanks a lot!

[aquilazhang]

    Hello,
    I suppose, the 'decoys_screening' you mean is the 'screening power' test? If so, there is no need to prepare small molecules again, the prepared small molecules I have uploaded can be directly used. The problematic molecules have been manually modified one-by-one to make sure they can be accepted by RDKit.By the way, the recently compiled CASF-2016 can be freely accessed at the lastest release of PDBbind (Demo samples at https://www.pdbbind-plus.org.cn/download). As far as I know, all molecules can be directly read by RDKit.

                    Zhang, XiangyingFudan ***@***.***

---- Replied Message ----

     From 

        ***@***.***>

     Date 

    4/19/2024 16:45

     To 

        ***@***.***>

     Cc 

        Xiangying ***@***.***>
        ,

        ***@***.***>

     Subject 

          Re: [ComputArtCMCG/PLANET] Tests on the CASF dataset (Issue #3)

Hello! Thanks for your interests in PLANET. In fact, the core set has been mannually checked and processed by myself. Since some mol structures in the currently availble CASF-2016 on the PDBbind website has some problems (maybe bond type etc.) Here, I will upload a processed version of core set on the main page of this repository, on which I tested PLANET and acheive the published result. I has just re-conducted the scoring power test and the result is the same as reported (in the attachment). CASF_test.zip

Hi, I would like to ask you how you dealt with small molecules in decoys_screening in CASF-2016 that could not be successfully read in by rdkit. I also encountered unreasonable bond issues when reading ligand molecules with rdkit, can you give me any advice, thanks a lot!

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: @.***>

[kongmenghua]

抱歉，刚刚才发现您的邮件，您在rdkit处理的时候具体遇到了什么问题呢？您这边方便加微信交流嘛，对于decoy的数据还有一些问题想跟您请教一下（我的微信是13620650781）

发件人：Xiangying Zhang @.> 发送日期：2024-04-19 17:03:39 收件人：ComputArtCMCG/PLANET @.> 抄送人：kongmenghua @.>,Author @.> 主题：Re: [ComputArtCMCG/PLANET] Tests on the CASF dataset (Issue #3)

Hello, I suppose, the 'decoys_screening' you mean is the 'screening power' test? If so, there is no need to prepare small molecules again, the prepared small molecules I have uploaded can be directly used. The problematic molecules have been manually modified one-by-one to make sure they can be accepted by RDKit.By the way, the recently compiled CASF-2016 can be freely accessed at the lastest release of PDBbind (Demo samples at https://www.pdbbind-plus.org.cn/download). As far as I know, all molecules can be directly read by RDKit.

Zhang, XiangyingFudan @.***

---- Replied Message ----

From

@.***>

Date

4/19/2024 16:45

To

@.***>

Cc

Xiangying @.***> ,

@.***>

Subject

Re: [ComputArtCMCG/PLANET] Tests on the CASF dataset (Issue #3)

Hello! Thanks for your interests in PLANET. In fact, the core set has been mannually checked and processed by myself. Since some mol structures in the currently availble CASF-2016 on the PDBbind website has some problems (maybe bond type etc.) Here, I will upload a processed version of core set on the main page of this repository, on which I tested PLANET and acheive the published result. I has just re-conducted the scoring power test and the result is the same as reported (in the attachment). CASF_test.zip

Hi, I would like to ask you how you dealt with small molecules in decoys_screening in CASF-2016 that could not be successfully read in by rdkit. I also encountered unreasonable bond issues when reading ligand molecules with rdkit, can you give me any advice, thanks a lot!

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: @.>— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.>