In deep learning based compound generation models the metric of fraction of RDKit-valid compounds is ubiquitous, but is problematic from the cheminformatics viewpoint as a large fraction may be driven by pyrrolic nitrogens (see below) rather than Texas carbons (carbon with 5 bonds like the Star of Texas). In RDKit, no error is more irksome that the KekulizeException or ValenceException from RDKit sanitisation. These are raised when the molecule is not correct. This would make the RDKit-valid a good metric, except for a small detail: the validity is as interpreted from the the stated implicit and explicit hydrogens and formal charges on the atoms, which most models do not assign. Therefore, a compound may not be RDKit-valid because it is actually impossible, like a Texas carbon, but in many cases it is because the formal charge or implicit hydrogen numbers of some atoms are incorrect. In both case, the major culprit is nitrogen. Herein I go through what they are and how to fix them, with a focus on aromatic nitrogens.
Tag Archives: RDKit
Tanimoto similarity of ECFPs with RDKit: Common pitfalls
A common measure for the similarity of two molecules is the Tanimoto similarity of their ECFPs (Extended Connectivity FingerPrint). However, there is no clear standard in literature for what kind of ECFPs should be used when calculating the Tanimoto similarity, and that choice can lead to substantially different results. In this post I wish to shed light on some results you should know about before you jump into your calculations.
A blog post on how ECFPs are generated was written by Marcus Dablander in 2022 so please take a look at that. In short, ECFPs have a hyperparameter called the radius r, and sometimes a fingerprint length L. Each entry in the fingerprint indicates the presence or absence of a particular substructure in the molecule of interest, and the radius r defines how large the substructures that you consider are. If you have r=3 then you consider substructures made by going up to three hops away from each atom in your molecule. This is best explained by this figure from Marcus’ post:
I really hope my compounds get the green light
As a cheminformatician in a drug discovery campaign or an algorithm developer making the perfect Figure 1, when one generates a list of compounds for a given target there is a deep desire that the compounds are well received by the reviewer, be it a med chemist on the team or a peer reviewer. This is despite scientific rigour and training and is due to the time invested. So to avoid the slightest shadow of med chem grey zone, here is a hopefully handy filter against common medchem grey-zone groups.
Continue readingComparing pose and affinity prediction methods for follow-up designs from fragments
In any task in the realm of virtual screening, there need to be many filters applied to a dataset of ligands to downselect the ‘best’ ones on a number of parameters to produce a manageable size. One popular filter is if a compound has a physical pose and good affinity as predicted by tools such as docking or energy minimisation. In my pipeline for downselecting elaborations of compounds proposed as fragment follow-ups, I calculate the pose and ΔΔG by energy minimizing the ligand with atom restraints to matching atoms in the fragment inspiration. I either use RDKit using its MMFF94 forcefield or PyRosetta using its ref2015 scorefunction, all made possible by the lovely tool Fragmenstein.
With RDKit as the minimizer the protein neighborhood around the ligand is fixed and placements take on average 21s whereas with PyRosetta placements, they take on average 238s (and I can run placements in parallel luckily). I would ideally like to use RDKit as the placement method since it is so fast and I would like to perform 500K within a few days but, I wanted to confirm that RDKit is ‘good enough’ compared to the slightly more rigorous tool PyRosetta (it allows residues to relax and samples more conformations with the longer runtime I think).
Mapping derivative compounds to parent hits
Whereas it is easy to say in a paper “Given the HT-Sequential-ITC results, 42 led to 113, a substituted decahydro-2,6-methanocyclopropa[f]indene”, it is frequently rather trickier algorithmically figure out which atoms map to which. In Fragmenstein, for the placement route, for example, a lot goes on behind the scenes, yet for some cases human provided mapping may be required. Here I discuss how to get the mapping from Fragmenstein and what goes on behind the scenes.
Continue readingFinding and testing a reaction SMARTS pattern for any reaction
Have you ever needed to find a reaction SMARTS pattern for a certain reaction but don’t have it already written out? Do you have a reaction SMARTS pattern but need to test it on a set of reactants and products to make sure it transforms them correctly and doesn’t allow for odd reactants to work? I recently did and I spent some time developing functions that can:
- Generate a reaction SMARTS for a reaction given two reactants, a product, and a reaction name.
- Check the reaction SMARTS on a list of reactants and products that have the same reaction name.
The workings of Fragmenstein’s RDKit neighbour-aware minimisation
Fragmenstein is a Python module that combine hits or position a derivative following given templates by being very strict in obeying them. This is done by creating a “monster”, a compound that has the atomic positions of the templates, which then reanimated by very strict energy minimisation. This is done in two steps, first in RDKit with an extracted frozen neighbourhood and then in PyRosetta within a flexible protein. The mapping for both combinations and placements are complicated, but I will focus here on a particular step the minimisation, primarily in answer to an enquiry, namely how does the RDKit minimisation work.
Pairwise sequence identity and Tanimoto similarity in PDBbind
In this post I will cover how to calculate sequence identity and Tanimoto similarity between any pairs of complexes in PDBbind 2020. I used RDKit in python for Tanimoto similarity and the MMseqs2 software for sequence identity calculations.
A few weeks back I wanted to cluster the protein-ligand complexes in PDBbind 2020, but to achieve this I first needed to precompute the sequence identity between all pairs sequences in PDBbind, and Tanimoto similarity between all pairs of ligands. PDBbind 2020 includes 19.443 complexes but there are much fewer distinct ligands and proteins than that. However, I kept things simple and calculated the similarities for all 19.443*19.443 pairs. Calculating the Tanimoto similarity is relatively easy thanks to the BulkTanimotoSimilarity function in RDKit. The following code should do the trick:
from rdkit.Chem import AllChem, MolFromMol2File
from rdkit.DataStructs import BulkTanimotoSimilarity
import numpy as np
import os
fps = []
for pdb in pdbs:
mol = MolFromMol2File(os.path.join('data', pdb, f'{pdb}_ligand.mol2'))
fps.append(AllChem.GetMorganFingerprint(mol, 3))
sims = []
for i in range(len(fps)):
sims.append(BulkTanimotoSimilarity(fps[i],fps))
arr = np.array(sims)
np.savez_compressed('data/tanimoto_similarity.npz', arr)
Sequence identity calculations in python with Biopandas turned out to be too slow for this amount of data so I used the ultra fast MMseqs2. The first step to running MMseqs2 is to create a .fasta file of all the sequences, which I call QUERY.fasta. This is what the first few lines look like:
Continue readingBRICS Decomposition in 3D
Inspired by this blog post by the lovely Kate, I’ve been doing some BRICS decomposing of molecules myself. Like the structure-based goblin that I am, though, I’ve been applying it to 3D structures of molecules, rather than using the smiles approach she detailed. I thought it may be helpful to share the code snippets I’ve been using for this: unsurprisingly, it can also be done with RDKit!
I’ll use the same example as in the original blog post, propranolol.
First, I import RDKit and load the ligand in question:
Continue reading
Molecular conformation generation with a DL-based force field
Deep learning (DL) methods in structural modelling are outcompeting force fields because they overcome the two main limitations to force fields methods – the prohibitively large search space for large systems and the limited accuracy of the description of the physics [4].
However, the two methods are also compatible. DL methods are helping to close the gap between the applications of force fields and ab initio methods [3]. The advantage of DL-based force fields is that the functional form does not have to be specified explicitly and much more accurate. Say goodbye to the 12-6 potential function.
In principle DL-based force fields can be applied anywhere where regular force fields have been applied, for example conformation generation [2]. The flip-side of DL-based methods commonly is poor generalization but it seems that force fields, when properly trained, generalize well. ANI trained on molecules with up to 8 heavy atoms is able to generalize to molecules with up to 54 atoms [1]. Excitingly for my research, ANI-2 [2] can replace UFF or MMFF as the energy minimization step for conformation generation in RDKit [5].
So let’s use Auto3D [2] to generated low energy conformations for the four molecules caffeine, Ibuprofen, an experimental hybrid peptide, and Imatinib:
CN1C=NC2=C1C(=O)N(C(=O)N2C)C CFF CC(C)Cc1ccc(cc1)C(C)C(O)=O IBP Cc1ccccc1CNC(=O)[C@@H]2C(SCN2C(=O)[C@H]([C@H](Cc3ccccc3)NC(=O)c4cccc(c4C)O)O)(C)C JE2 Cc1ccc(cc1Nc2nccc(n2)c3cccnc3)NC(=O)c4ccc(cc4)CN5CCN(CC5)C STIContinue reading