I was at the recent RDKit User Group meeting. One of the issues was long-term funding. Another was leadership succession.
The video shows Google's project is heavily dependent on RDKit. How much funding are they willing to contribute to RDKit?
I am not a pharmaceutical chemist. I am a nitpicking cheminformatics software developer.
At 2:07 it shows the cohort "Lorlatinib_Macrocycle" has a topology of "Macrocyclic (12 ≤ ring ≤ 16)" but the Key Structure Hallmarks are "12- to 15-member macrocyclic rings". Why is one 16 and the other 15?
I assume J Med Chem 2014 is https://pubs.acs.org/jmcmar/article-abstract/57/11/4720/1301... . I can see table 1, "Potency and ADME of Compound 1 and Other Key Acyclic Leads" which has the 6a-6i, for example. I think there's a mistake at 2:10 for the molecular formula of Compound 7a (Ether Macrocycle). The video says it's C19 but I only see 18 carbons. I looked at Scheme 1, and again only see 18 carbons (note the n=0 for compound 7a). I think it's supposed to be 7b?
At 3:22 it mentions "Chemical Similarity (Tanimoto ECFP4) then later at 3:37 it uses Morgan fingerprints. Are these generated using two different tools? While RDKit's Morgan r=2 is conceptional identical to ECFP4, saying "ECFP" sometimes implies using ChemAxon, or perhaps CDK, which use "ECFP" + diameter rather than RDKit's "Morgan" + radius. The latter timestamp doesn't specify the Morgan radius, which is important. (I'll assume also that it's using RDKit's binary Morgan fingerprints, and not the less commonly used count Morgan fingerprints, since that's the usual practice.)
At 3:27 they describe their matched molecular pairs. What approach do they use? I wrote most of the MMP package "mmpdb", building on previous work by Hussain and Rea, but there are others. The example shows "Acyclic-to-Macrocyclic Ring Closure Pairs", which is not something my code could handle.
At 3:29 they mention "MMPA analysis". Since the "A" in "MMPA" means "A"nalyis, that's like saying PIN number. ;)
I was at the recent RDKit User Group meeting. One of the issues was long-term funding. Another was leadership succession.
The video shows Google's project is heavily dependent on RDKit. How much funding are they willing to contribute to RDKit?
I am not a pharmaceutical chemist. I am a nitpicking cheminformatics software developer.
At 2:07 it shows the cohort "Lorlatinib_Macrocycle" has a topology of "Macrocyclic (12 ≤ ring ≤ 16)" but the Key Structure Hallmarks are "12- to 15-member macrocyclic rings". Why is one 16 and the other 15?
I assume J Med Chem 2014 is https://pubs.acs.org/jmcmar/article-abstract/57/11/4720/1301... . I can see table 1, "Potency and ADME of Compound 1 and Other Key Acyclic Leads" which has the 6a-6i, for example. I think there's a mistake at 2:10 for the molecular formula of Compound 7a (Ether Macrocycle). The video says it's C19 but I only see 18 carbons. I looked at Scheme 1, and again only see 18 carbons (note the n=0 for compound 7a). I think it's supposed to be 7b?
There is it again, at 2:41 showing that compound 7a is CHEMBL3286814 which https://www.ebi.ac.uk/chembl/explore/compound/CHEMBL3286814 says is indeed C18, not C19.
At 3:22 it mentions "Chemical Similarity (Tanimoto ECFP4) then later at 3:37 it uses Morgan fingerprints. Are these generated using two different tools? While RDKit's Morgan r=2 is conceptional identical to ECFP4, saying "ECFP" sometimes implies using ChemAxon, or perhaps CDK, which use "ECFP" + diameter rather than RDKit's "Morgan" + radius. The latter timestamp doesn't specify the Morgan radius, which is important. (I'll assume also that it's using RDKit's binary Morgan fingerprints, and not the less commonly used count Morgan fingerprints, since that's the usual practice.)
At 3:27 they describe their matched molecular pairs. What approach do they use? I wrote most of the MMP package "mmpdb", building on previous work by Hussain and Rea, but there are others. The example shows "Acyclic-to-Macrocyclic Ring Closure Pairs", which is not something my code could handle.
At 3:29 they mention "MMPA analysis". Since the "A" in "MMPA" means "A"nalyis, that's like saying PIN number. ;)