Review of Symposium ‘Innovative Software for Molecule Discovery and Design’, New Delhi, India

Manoranjan Singh Sidhu, Neotel Systems & Services (Cresset distributor, India)

On 12th April 2019, the International Centre for Genetic Engineering and Biotechnology, New Delhi, India, hosted the symposium ‘Innovative Software for Molecule Discovery and Design’. Delegates learnt how experienced users at the Institute had used Cresset computational tools to efficiently discover better compounds.

Dr Robert Scoffin, Cresset CEO, opened the symposium with an explanation of Cresset’s patented XED force field. He spoke in detail about the ligand-based and structure-based applications, emphasizing how Cresset technology helps researchers with better visualization and ease of use. Dr Scoffin explained how, Flare™, a new application enabling enhanced designs by using new approaches to protein-ligand analysis, can streamline new molecule design using Electrostatic Complementarity™; this provides rapid activity prediction with visual feedback on new molecule designs and proves invaluable for understanding ligand binding, structure-activity relationships and for ranking new molecule designs.

In a demonstration of Spark™, Cresset’s scaffold hopping and R-group exploration application, Dr Scoffin showed how it can be used to generate highly innovative ideas for your project to escape IP and toxicity traps. Dr Scoffin explained how Spark gives a single assessment of 20 different datasets, thus providing greater insight and adding greater value when compared with alternative applications that build libraries from larger datasets.

Dr Suneel Kumar, Cresset Application Scientist, demonstrated SAR analysis using the Activity Miner™ and Activity Atlas™ components of Forge™, showing how these modules are useful in understanding SAR of the current dataset and how they provide insights to design better molecules.

The symposium was very interactive with many delegates asking questions regarding visualization and scoring correlations, how Cresset software can help fill gaps or complement their existing infrastructure, how large datasets could be reduced to a lesser number of assessments so as to understand the results better, and how Cresset technology could help with synthetic feasibility and commercial availability of a lead molecule.

I encourage anyone who is interested in learning more about how Cresset applications can advance their molecule design projects to request a free evaluation of ligand-based applications or structure-based applications and subscribe to the Cresset newsletter.

“As per the presentation and demonstration, the software provides excellent visualization and gives reliable results, and we look forward to evaluating Forge, Spark and Flare.”

Bioinformatics Group, International Centre for Genetic Engineering and Biotechnology, New Delhi, India

Application of Spark to the discovery of novel LpxC inhibitors

Novartis Institutes for Biomedical Research recently published the paper Application of Virtual Screening to the Identification of New LpxC Inhibitor Chemotypes, Oxazolidinone and Isoxazoline. They report using Spark™, Cresset’s scaffold hopping application, to find core replacements for the indazole moiety of compound 6 in Figure 1. Visual selection of the most promising top-ranking Spark results and further optimization studies led to the identification of novel LpxC inhibitors with subnanomolar binding to LpxC and in vivo antibacterial activity against P. aeruginosa and other Gram-negative bacteria.

Scaffold hopping

The bioactive conformation of compound 26, a simplified version of compound 6 originally reported by Actelion, was used as the starter molecule for the Spark scaffold hopping experiment. Aim of the search was to identify appropriate replacements of the indazole core (Figure 1).


Figure 1. Top: LpxC inhibitor 6. Bottom: The bioactive conformation of compound 26 was used as the starter molecule for the scaffold hopping experiment with Spark.

The allowed atom types for the ‘linker 1 atom’ and ‘atom 2’ (see Figure 2) were set respectively to ‘any carbon atom’ and ‘any atom’. It was further specified that all Spark designs must contain at least 1 ring and should not include any reactive functionalities. The Spark experiment was performed on fragment databases derived from ZINC,1 ChEMBL,2 and the VEHICLe3 collection of theoretical ring systems.

The similarity score of the Spark results towards compound 26 was calculated using 50% field and 50% shape similarity. The 100 top ranking clusters were manually reviewed with respect to synthetic feasibility, introduction of hydrophilic groups in the area of the indazole moiety and calculated physicochemical properties.


Figure 2. Allowed atom types for ‘linker 1 atom’ and ‘atom 2’  in the Spark experiment.

The oxazolidinone 13 and isoxazoline 25 scaffolds were shortlisted among several proposals which link ‘linker 1 atom’ and ‘atom 2’ with an hydrophilic linker (Figure 3).

Analogues of 13 and 25, where the methoxy group in para position was replaced by a bromine atom (Figure 3) showed cellular activity with Minimal Inhibitory Concentrations (MIC) values below 4 μg/mL against P. aeruginosa. Accordingly, these series were selected for further investigation.


Figure 3. Oxazolidinone 13a and isoxazoline 25a showed  MIC <4 μg/mL against P. aeruginosa.

SAR optimization

Further investigation and expansion of both oxazolidinone and isoxazoline series led to the identification of compounds with potent in vitro activity against P. aeruginosa and other Gram-negative bacteria. Representative compound 13f (Figure 4) demonstrated excellent efficacy against P. aeruginosa in an in vivo mouse neutropenic thigh infection model.

The crystal structure of 13f complexed with the P. aeruginosa LpxC enzyme (PDB: 6MAE) shows that the hydroxamic acid moiety is bound to the zinc atom in the active site, and involved in interactions with H78, H237, T190, E77, and D241 (Figure 4). The hydrophobic tail (phenyl group) interacts with several hydrophobic side chains, and the cyclopropyl group further extends to the solvent exposed region. The sulfone oxygen atoms interacts with well-defined crystallographic water molecules (not shown in Figure 4), and the methyl group attached to the sulfone functionality is engaged in hydrophobic interactions with F191. The carbonyl oxygen of the oxazolidinone forms a favorable polar interaction with the C2 CH group of H19.

The crystal structure of 13f thus confirms that the oxazolidinone group can favourably orient the sulfone/hydroxamic acid portion of the inhibitor with respect to the hydrophobic phenyl group while maintaining a low-energy conformation, as predicted by Spark.

Figure 4. X-ray crystal structure of 13f complexed with P. aeruginosa LpxC enzyme (PDB: 6MAE)



This paper from NIBR shows that Spark can guide the rational design and discovery of new drug candidates. In this case, a traditional scaffold hopping experiment led to the identification of novel oxazolidinone and isoxazoline LpxC inhibitors with potent antibacterial activity against Gram-negative bacteria.

Try Spark on your project

Request a free evaluation of Spark to move to new series and non-obvious IP by swapping scaffolds.


  1. Irwin, J. J.; Sterling, T.; Mysinger, M. M.; Bolstad, E. S.; Coleman, R. G. ZINC: A Free Tool to Discover Chemistry for Biology. J. Chem. Inf. Model. 2012, 52 (7), 1757–1768.
  2. Gaulton, A.; Bellis, L. J.; Bento, A. P.; Chambers, J.; Davies, M.; Hersey, A.; Light, Y.; McGlinchey, S.; Michalovich, D.; Al-Lazikani, B.; Overington, J. P. ChEMBL: A Large- Scale Bioactivity Database for Drug Discovery. Nucleic Acids Res. 2012, 40 (D1), D1100–D1107.
  3. Pitt, W. R.; Parry, D. M.; Perry, B. G.; Groom, C. R. Heteroaromatic Rings of the Future. J. Med. Chem. 2009, 52 (9), 2952–2963

Application of Spark in the discovery of potent SOS1 inhibitors that block RAS activation via disruption of the RAS–SOS1 interaction

Bayer AG recently published an interesting paper on the discovery of potent and selective SOS1 inhibitors that block RAS activation via disruption of the RAS–SOS1 interaction.

They report using Spark™, Cresset’s bioisostere replacement tool, to rationally design linkers between active hit compounds from fragment screen and HTS. Herein, we review this interesting application of Spark to a ligand joining experiment that led to the discovery of BAY-293, a potent and selective inhibitor of KRAS-SOS1 interaction.

Fragment screen

Bayer decided on a two-pronged approach, running a HTS campaign and a fragment screen in parallel. The fragment screen, with a library of 3,000 fragments, led to the identification of fragments which bind to and can induce a conformational change at the KRAS-SOS1 protein-protein interaction site. By triggering a rotation of the Phe890 side chain, they open a new subpocket adjacent to the main binding pocket.

F1 was chosen as the starting point for further optimization. The crystal structure of F1 in complex with SOS1 (PDB: 6EPM, Figure 1) shows that the phenyl moiety makes a π–π interaction with Phe890 (Phe-out). The aminomethyl moiety forms hydrogen bonds to Asp887 and the backbone carbonyl of Tyr884 and makes an additional cation–π interaction with the Tyr884 side chain.

Figure 1. X-ray crystal structure of F1 complexed with KRASG12C–SOS1cat (PDB: 6EPM).

HTS and initial optimization

HTS of a Bayer library of 3 million compounds led to the identification of compound 1 (IC50 320 nM, Figure 2).

Figure 2. Compound 1 (HTS hit: IC50 320 nM).

Replacement of the naphthyl moiety in compound 1 with a pyrazolylphenyl group resulted in compound 17 (Figure 3), showing a good potency on SOS1 (IC50 140 nM) and improved aqueous solubility.

In terms of interactions with SOS1 (PDB: 5OVF, Figure 3), the quinazoline scaffold of compound 17 is sandwiched between His905 and Tyr884 (π–π stacking). The pyrazolylphenyl moiety occupies a hydrophobic pocket composed of Leu901 and Phe890 (Phe-in) and makes a T-stacking interaction with the side chain of Tyr884. The pyrazole moiety makes a water-bridged H-bond to Glu902. The central aniline NH makes a H-bond with the side chain of Asn879.

Figure 3: X-ray crystal structure of compound 17 in the SOS1SB active site (PDB: 5OVF).

Linking of the fragment and HTS hits with Spark

As the fragment screen identified a new subpocket that was not yet addressed by the HTS hits, scientists at Bayer attempted a ligand joining approach using Spark to try and further improve potency by combining both ligand series. A superimposition of the crystal structures of compound 17 (PDB: 5OVF) and F1 (PDB: 6EPM), shown in Figure 4 – A, suggests that this approach is feasible.

After cutting out the overlapping aryl groups of compounds 17 and F1, the authors performed ligand joining experiments with Spark (Figure 4 – B, C) to identify appropriate linkers that could correctly orientate both the tetrahydrocyclopenta[c]pyrazole moiety of F1 ­­­­and the aminoquinazoline group of 17 within the active site of SOS1. Among the top-scoring linkers suggested by Spark, the thiophene linker was chosen, synthesized, and found to be active.

Figure 4: Ligand-joining approach of fragment and HTS hits. A) Superimposition of the crystal structures of F1 bound to KRASG12C_SB–SOS1cat (SOS1 in gray, F1 and Phe890 with carbon atoms in magenta) and 17 bound to SOS1SB (17 and Phe890 shown; carbon atoms in green). B) Schematic depiction of the merging approach. C):  Representation of F1 and 17 with respective fields from Cresset.

Further optimization and discovery of BAY-293

Optimization of the hybrid hits containing the thiophene linker identified by Spark by introducing polar moieties (such as OH, NH2) that mimic the hydrogen bond interactions of the amino side chain of F1, and trigger the Phe-out conformation, led to the discovery of the optimized compound 23 (BAY-293, IC50 21nM). As shown in Figure 5, the side chain amino group of compound 23 interacts with SOS1 by making H-bond interactions with Asp887 and Tyr884 backbone carbonyl, as well as a cation-π interaction with the Tyr884 side chain.

Figure 5. X-ray crystal structure of compound 23 in the SOS1 active site (PDB: 5OVI).

Further screening and antiproliferative data proved that compound 23 (BAY-293) is a potent and selective inhibitor of KRAS-SOS1 interaction and indicates that inhibition of GEFs may represent a viable approach for targeting RAS-driven tumors.


This paper by Bayer shows the utility of the Spark approach. As well as being a superb bioisostere finder, Spark’s advanced capabilities include easy-to-use methods for water replacement, macrocyclization, fragment growing and fragment linking. In this case Spark was vital in suggesting how to join two disparate molecules occupying different parts of a protein binding site, transferring SAR from one series to another.

Try Spark on your project

Request a free evaluation of Spark to generate highly innovative ideas for your project.

Which macrocycle should I try first? Picking the best linkers with Flare™ and Spark™

At Cresset, we enjoy seeing our products work in synergy. By combining the most recent scientific methods and workflows we deliver solutions to address molecule design challenges. In this post, we use the new Electrostatic Complementarity™ (EC) maps and scores in Flare to help the post-processing of a Spark macrocyclization experiment.

Using Electrostatic Complementarity in Flare to post-process the Spark results

In the case study Using Spark to design macrocycle BRD4 inhibitors, we used Spark, our bioisostere replacement and scaffold hopping tool, to design macrocyclization strategies for non-macrocyclic, pyridone BRD4 inhibitors and evaluate results against experimental data reported by Wang et al [1]. The results showed that Spark successfully reproduced the experimental data.

In a real drug discovery project where no retrospective data is available, it would be useful to have criteria based on the existing knowledge of the system under study helping a further post-processing of the Spark results. Here we show how to use Spark in synergy with the EC maps and scores in Flare, our structure-based design tool, to pick the most promising candidates for synthesis.

Electrostatic interactions are essential for molecular recognition and are also key contributors to the binding free energy ΔG of protein-ligand complexes. Assessing the electrostatic match between ligands and binding pockets provides important insights into why ligands bind and what can be changed to improve binding.

The 100 top scoring results from the BRD4 Spark experiment were opened in Flare using the ‘Send to Flare’ functionality in Spark, which also transfers the related starter molecule (compound 1 in Figure 1) and excluded volume protein (5UEY). The protein was prepared in Flare, removing the water molecules that do not make clear interactions with both the ligand and protein. EC scores and maps were then calculated for compound 1 and the experimentally validated macrocycle 2 reported by Wang et al. towards the same 5UEY protein, as shown in Figure 1. As expected, the EC maps for both compounds show good complementarity to the protein and a very similar EC R score of 0.52/0.53 (Pearson’s r correlation coefficient). Spark linkers showing similar (or better) maps/score should provide interesting ideas for synthesis.

Figure 1: EC maps and scores for compound 1 and macrocycle 2, calculated towards protein 5UEY. Color coding: green = good complementarity; red = electrostatic clash.

Picking the winners

Figure 2 shows a couple of the most interesting linkers in terms of EC score.

Figure 2: EC maps and scores (top panel) for two ‘matching’ Spark linkers, calculated towards protein 5UEY. Color coding: green = good complementarity; red = electrostatic clash. The bottom panel shows electrostatic potential maps for the same Spark results. Color coding: cyan = negative electrostatic; red = positive electrostatic.

In the first example (Figure 2 – left), the π-system in the double bond linker complements the positive electrostatic field at the NH proton of His437 better than compound 1 or a fully saturated linker of similar length, as in macrocycle 2.

Another interesting example of good electrostatic match is the mercaptoethanol linker (Figure 2 – right). The negative electrostatic field of the thioether group is also in close proximity to the polarized NH of His437.

For both compounds, the increase in EC towards the protein is due to the introduction of a more negative ligand electrostatic in the region near His437, as shown by the electrostatic potential maps for both linkers (Figure 2 – bottom).

Discarding the losers

In contrast, an analysis of the EC maps for two of the linkers with the lowest EC scores (Figure 3) immediately highlights the reasons why these should be down-prioritized.

Figure 3. EC maps and scores (top panel) for two ‘clashing’ Spark linkers, calculated towards protein 5UEY. Color coding: green = good complementarity; red = electrostatic clash. The bottom panel shows electrostatic potential maps for the same Spark results. Color coding: cyan = negative electrostatic; red = positive electrostatic.
These linkers expose an area of negative interaction potential towards the carbonyl of Asn443, resulting in a strong electrostatic clash.


Are you surprised that a few linkers with low EC ended up among the top 100 scoring Spark results? Don’t forget that Spark works on ligand similarity. In macrocyclization (and fragment linking) experiments we are stretching the method to explore regions in space where ‘no ligand has gone before’.

In such cases, adding protein information is clearly highly beneficial to help post-processing. EC maps in Flare are an intuitive visual method for rationalizing the choice of the best ideas to progress, while EC scores provide a rapid way of scoring and filtering the 500 Spark results in just a few minutes.

To try Spark or Flare on your projects, request your free evaluation.

  1. Wang, L.; McDaniel, K. F.; Kati, W. M. Fragment-Based, Structure-Enabled Discovery of Novel Pyridones and Pyridone Macrocycles as Potent Bromodomain and Extra-Terminal Domain (BET) Family Bromodomain Inhibitors. J. Med. Chem. 2017, 60 (9), 3828–3850.

Rapid and accessible in silico macrocycle design


Macrocyclization of pharmaceutical compounds plays an increasing role in drug discovery. Macrocycles can provide several advantages such as favorable drug-like properties, increased selectivity and improved binding affinity.

The assessment of potential macrocyclizations is a challenging computational problem. Linkers should be synthetically feasible, not too flexible, have a sensible conformation and be pharmacophorically compatible with the active site. Few computational strategies are available for this task, so most design is done ad hoc.

Here we present a modification of our existing bioisostere searching tool, Spark, to this problem. Traditional bioisostere searches specify a fragment to replace in the starting ligand, and look for similar fragments in a database. Instead, Spark assesses bioisosteric replacements in product space, which allows more complex experiments. A modification of the Spark scoring method to include similarity to other ligands known to bind in the region of the linker, as well as constraints from the protein active site and known pharmacophoric requirements, allows sensible ranking of potential linkers.

Here we present case studies of designing macrocyclization strategies for BRD4 and for Orexin 2. The Spark algorithms enable a rapid assessment of the ideal linker length and suggested chemistry for each cyclization option.

See the ‘Rapid and accessible in silico macrocycle design’ presentation Rob Scoffin presented in the COMP Division at the 256th ACS National Meeting.

Outstanding new 3D graphics in Spark 10.5.5

A new patch level release of Spark™, our scaffold hopping and bioisostere replacement application, is now available for download by all Spark users. Spark 10.5.5 includes considerable improvements to the look and feel, rendering and performance of the graphics of the 3D window.

Figure 1. Improved 3D graphics in Spark 10.5.5.

Spark 10.5.5 also includes a small number of additional improvements and bug fixes:

  • Improved support for the configuration of proxy servers
  • Improved Spark start up times when using databases sitting in a remote location on a slow connection
  • Improved support for high-dpi displays
  • Fixed issue which caused the effect of the application of pharmacophore constraints to be overestimated in some circumstances
  • Fixed issue on macOS which prevented to dock back in the desired position any dock window moved outside of the main Spark interface
  • Fixed rare issue in the wizard where in some occasions the desired hydrogen atom could not be picked for replacement.

Download Spark 10.5.5

To ensure you benefit from the improved 3D graphics, and other improvements and bug fixes, keep an eye out for an email with download links and upgrade Spark at your earliest convenience.

If you are not currently a Spark customer, please request a free evaluation.

Contact us if you have queries.

Pharmacophore constraints – when should they be used?

Cresset’s alignment and scoring technology is based around the idea of electrostatic matching: the ‘best’ alignment of a pair of conformations is generally the one that provides the most similar molecular interaction potentials. In most cases this is true. However, we recognize that often some portions of the electrostatic potential, for example those involved in a strong hydrogen bond, can be more important than others. This isn’t information intrinsic to the molecules themselves, but to the context in which they are being used. As a result, we have for many years offered the possibility of adding ‘field constraints’, which force the alignment algorithm to prioritize matching the electrostatic potential in particular places around a ligand.

It’s worth noting that the field constraint algorithm works by applying a score penalty to alignments which violate the constraint, not by forcing the constraint to be matched (Figure 1). In general, soft constraints delivered via score penalties are preferred to hard constraints where we simply disallow alignments that violate the constraint. There are two reasons behind this. Firstly, using a score penalty allows the strength of the constraint to be adjusted, so that it can be anything from a mild hint to a strict requirement. Second, in most cases we feel that it is better to get a best-effort alignment that violates the constraint than nothing at all, especially when using the algorithms in lead optimization and trying to align a set of known actives.

Figure 1: Field constraint requires the field at the position of a particular field point to have a certain minimum value.

While field constraints work extremely well in the majority of cases, there are a few exceptions. In some cases there isn’t a field point at the place you want to add a constraint (although this problem is alleviated by the new field editor in Forge and Spark). More commonly, the constraint that you want to add is either stronger or more specific than can be expressed through fields: sometimes generic positive potential isn’t enough and you want to explicitly require a hydrogen-bond donor, for example, and sometimes you want to ensure that the metal-chelating group in a set of ligands is aligned, and ‘good metal chelator’ is similarly difficult to pin down in terms of pure electrostatics. In addition, the donor-acceptor motif that is often required in the hinge binding region of kinases can be difficult to describe through fields alone, as the donor and acceptor electrostatics cancel out to some extent making this region contribute less to the overall alignment score than it should.

For these reasons we introduced the option of pharmacophoric constraints to our core algorithms. These operate in a similar fashion to field constraints, in that they are constraints not requirements. Traditional pharmacophore algorithms treat pharmacophores as binary switches: you either have a H-bond donor in a region of space, and hence a full match, or you don’t and therefore don’t have a match. Our pharmacophore constraints instead operate on a sliding scale. You specify how strong the constraint is, and hence how much alignment score should be lost if the constraint is not satisfied. We spent a while looking at what form the penalty function should take, and the best form overall was a simple linear penalty with a cap at 2Å distance (Figure 2).

Figure 2: Pharmacophore constraints require an atom of the specified type to be within 2Å of the constrained atom, or a penalty is applied to the score.

We’ve talked elsewhere about the validation of our pharmacophore constraints on virtual screening performance within Blaze, where for some targets they can make a significant difference. However, the new constraints are useful for more than just virtual screening. What effect does adding constraints have on the probability of a successful ligand alignment?

We investigated this by using the standard AZ/CCDC dataset from Giangreco et al. (J. Chem. Inf. Model. 2013, 53, 852−866), which contains a large number of targets, and for each target a collection of aligned co-crystallized ligands. The data set can be used to validate ligand alignment algorithms, by measuring what proportion of the ligands within a target class can be successfully aligned. However, running a validation over the entire data set would be challenging, as it would pose the question for each target as to what pharmacophore constraints should be applied. We decided to address this by only testing (in the first instance) 4 kinase targets, as the donor-acceptor-donor motif for kinases is well understood and fairly uncontroversial. For each ligand in these 4 target sets, we manually applied between 1 and 3 pharmacophore constraints to the hinge binding region based on how many H-bonds were actually made by that ligand. For the purposes of this experiment we only included ‘true’ H-bonds from hydrogens on O or N, not the weaker C-H form. Note that all of the alignments were performed in ligand space only, with no information from the proteins applied.

The distribution of results is shown in Figure 3. We used a cutoff of 2Å to distinguish between correct and incorrect alignments. As can be seen, a significant number (around 13%) of alignments that were incorrect when unconstrained were rescued by adding the constraints – that’s a significant improvement. However, at the same time, around 3% of the alignments that we had been getting correct were made worse by adding the constraints, and it’s also apparent that there are a number of incorrect alignments which are left unchanged by the imposition of constraints.  Overall the pharmacophores are a net win, especially for these kinase data sets, but it’s not a magic bullet.

Figure 3: Effect of adding pharmacophore constraints to 4 kinase alignment data sets. The alignments in the ‘Win’ section are improved by the addition of constraints, while those in the ‘Loss’ section are made incorrect.

Let’s look at a couple of examples. Figure 4 shows a case where adding hinge binder constraints converts an incorrect alignment to a correct one. The unconstrained result on the left matches the dimethoxyphenyl ring on the left and the phenol on the right very well, but at the expense of the quinazoline nitrogen which makes the hinge interaction. Adding a single acceptor constraint to that nitrogen switches the best solution to one where the indazole nitrogen satisfies the constraint, which happens to be the experimentally-observed alignment.

Figure 4: CDK2, aligning EZV to DTQ.

Similarly, in Figure 5, the best alignment in terms of raw shape and electrostatics of MTW and LZ8 is shown on the left. The 2-aminopyrimidine in MTW which makes the hinge interactions is partially matched by LZ8’s pyrazole and carbonyl, but the alignment in that part of the molecule isn’t that good. However, the alignment matches the phenyl ring in both molecules beautifully. Applying a donor/acceptor pharmacophore constraint pair forces the alignment algorithm to try and match the 2-aminopyrimidine much more closely, which leads to the correct alignment on the right. However, note that the phenyl rings are much more poorly aligned.

Figure 5: CDK2, aligning LZ8 to MTW.

Finally, Figure 6 shows a case where adding field constraints converts a correct alignment to an incorrect one. When aligning to DTQ as the reference, as in Figure 4, the field/shape alignment gets the correct alignment for C94. However, C94 does not make the hydrogen bond to the hinge that we have constrained in the reference. The constraint coerces the alignment to be flipped so that one of the sulfonamide oxygens matches the H-bond acceptor requirement, thus forcing the wrong answer to be obtained.

Figure 6: CDK2, aligning C94 to DTQ.

Pharmacophore constraints can, as you can see, be very useful. If you know that particular interactions are required by the protein, you can provide our tools with that information to help them find the correct alignment. However, it must be kept in mind that by adding these constraints you are adding bias to the experiment: you are nudging it to provide the answers that you expect. Sometimes, the unexpected answers turn out to be correct! We recommend, therefore, that pharmacophore constraints are used sparingly. I personally will always run an unconstrained alignment experiment as a control, just to see if Forge will find an alignment that I hadn’t expected but which might illuminate a previously-puzzling piece of SAR.

So, when should you use a field constraint and when a pharmacophore constraint? The answer depends both on your requirements and on the chemistry inside the active site. If there’s a pharmacophoric interaction that you are certain must be preserved, then using a pharmacophore constraint is probably justified. However, if the interaction is not always present in the same form, a field constraint may be more appropriate. Figure 7 shows a BTK example. The ligand on the left makes two hydrogen bonds to the backbone from the indazole. We might be tempted to add pharmacophore constraints to both the donor and the acceptor, except that there are known highly-active ligands which make a C-H hydrogen bond instead (see the ligand on the right). As a result, it is probably more appropriate, if constraints are required, to add a pharmacophore constraint to the acceptor nitrogen, but use a field constraint to indicate a donor preference in the NH/CH location.

Figure 7: BTK ligands – which constraint type should I use?

Try yourself

Pharmacophore constraints are now available in all Cresset ligand-based computational chemistry applications:

  • Forge: Powerful ligand-focused suite for understanding SAR and design
  • Torch: Molecular design and SAR for medicinal and synthetic chemists
  • Spark: Discover new directions for your project using bioisosteres
  • Blaze: Effective virtual screening optimized to return diverse new structures

Request an evaluation and try them yourself.

New release of Spark databases

A new release of the Spark™ fragment and reagent databases is now available for download, to accompany the release of Spark 10.5. These are designed to provide you with an excellent source of new biososteres, whilst also ensuring that the results of your Spark experiment are tethered to molecules which are readily synthetically accessible.

Fragment Databases

In this release we have made significant additions to the source of our fragment databases. The new Spark ‘Commercial’ databases (replacing the previous ZINC fragments) use the combination of ZINC15 and the eMolecules Screening Compounds and include significant new chemical diversity.

The Spark ‘ChEMBL’ databases have also been updated and are based on release 23 of ChEMBL.

In all cases, the compounds in the entire source collection were filtered to remove potentially toxic or reactive fragments. They were then fragmented, and the frequency with which any fragment appeared in the original source database annotated. The fragments were then sorted by frequency and labelled according to the number of bonds that were broken to obtain the fragment, as shown in the table below.

Spark Category Database Total number of fragments (to nearest 1000) Frequency
Commercial VeryCommon 64,000 Fragments which appear in more than 650 molecules
Common 137,000 Fragments which appear in 140-649 molecules
LessCommon 256,000 Fragments which appear in 35-139 molecules
Rare 401,000 Fragments which appear in 12-34 molecules
VeryRare 675,000 Fragments which appear in 5-11 molecules
ExtremelyRare 749,000 Fragments which appear in 3-4 molecules
ChEMBL Common 306,000 Fragments which appear in more than 6 molecules
Rare 506,000 Fragments which appear in 2-6 molecules
Very rare 570,000 Fragments which appear in a single molecule


Overall, the new Spark databases include over 3 Million fragments which can be used to identify novel bioisosteres for your project. Figure 1 plots the number of fragments in each database per connection point count.

Figure 1: Count of fragments in Spark ‘Commercial’ (from ZINC15 and eMolecules’ Screening Compounds) and ‘ChEMBL’ (from ChEMBL23) databases split by the number of connection points of each fragment.An analysis of the numbers of fragments in common between the ‘Commercial’ and ‘ChEMBL’ Spark databases (expressed as percent overlap with respect to ChEMBL) reveals that the databases overall show an excellent level of complementarity.

% overlap with ChEMBL Very Common Common Less Common Rare Very Rare Extremely Rare
ChEMBL common 16% 18% 15% 10% 8% 4%
ChEMBL rare 1% 5% 9% 10% 10% 7%
ChEMBL very rare 0% 2% 4% 6% 7% 5%

Not surprisingly, the most common fragments for each database significant overlap. However, the majority of ‘rare’ fragments appear to be unique to each database, showing that the original ZINC plus eMolecules’ Screening Compounds and the ChEMBL collections occupy quite distinct parts of chemical space.

Reagent databases

Monthly updates of the Spark reagent databases, derived from the eMolecules building blocks using an enhanced set of rules for chemical transformation, will continue also in this release. The February edition includes over 500,000 reagents with up-to-date availability information, to make it easy for you to move from the results of a Spark experiment to ordering the reagents you require to turn these results into reality.

The number of fragments in each reagent database is plotted in Figure 2.

Figure 2: Number of fragments in the Spark eMolecules reagent databases.
Each fragment in the eMolecules database is linked back to both the eMolecules ID for the source reagent and its availability. The advanced filtering capabilities in Spark (Figure 3) make it very easy to choose the optimal set of reagents for your experiment based on the Spark similarity score, preferred chemistry (as encoded by the reagent database which generated the result), availability information and overall physico-chemical profile of the results molecules.

Figure 3: Spark reagent results include availability information from eMolecules.
The eMolecules IDs for the favorite reagents can be easily exported from Spark and used to purchase the compounds from the eMolecules building blocks database, as shown in the web clip How to use the eMolecules reagents databases in Spark.

Create your own database

Spark fragment and reagent databases provide an excellent source of new bioisosteres. However, if you have access to significant proprietary chemistry, to specialized reagents, or simply want to only consider fragments from reagents that you have in stock then the creation of custom databases will add value to your Spark experiments.

The Spark Database Generator is a user-friendly interface within Spark that lets you easily create custom databases.

Figure 4: The Spark Database Generator.


This release of the fragment databases significantly increases the chemical diversity available to Spark users, while the monthly updates of the reagent databases ensure that the results of your Spark experiment are tethered to molecules which are readily synthetically accessible.

We are confident that these new and updated Spark fragment and reagent databases, combined with databases from your corporate collections generated with the Spark Database Generator, will provide an even better range bioisosteres for your project.

Please contact us to update to the latest databases, if you wish to access the Spark Database Generator, or to find out how Spark can impact your project.

Using Spark to design macrocycle BRD4 inhibitors


Macrocyclization of pharmaceutical compounds plays an increasing role in drug discovery. Macrocycles can provide several advantages such as favorable drug-like properties, and increased selectivity and binding affinity. Here we present a case study of designing macrocyclization strategies for reported BRD4 inhibitors with Spark1, Cresset’s bioisostere replacement and scaffold hopping tool.


Macrocyclization of pharmaceutical compounds can provide several advantages, such as diverse functionality, favorable drug-like properties, and increased selectivity.2 Also improved binding can result from macrocyclization by locking the molecule in a low-energy binding conformation and reducing entropic cost upon binding. 2–4 One challenge in harnessing the advantages of macrocyclic compounds is the difficulty in synthesizing such molecules. Prediction of effective macrocyclization strategies prior to synthesis is, therefore, crucial for successful macrocycle drug discovery.

Recently, Wang et al. reported a new class of pyridone-based BRD4 bromodomain inhibitors that show promising anticancer effects in cell lines and xenograft models.4 Starting from a pyridone fragment hit, the compound was further optimized using structure-based design to achieve one-digit nanomolar potency. Macrocyclization further improved binding affinity, cellular efficacy, and pharmacokinetic compound properties (Figure 1). 4

In this case study, we used Spark, Cresset’s bioisostere replacement and scaffold hopping tool, to design macrocyclization strategies for non-macrocyclic, pyridone BRD4 inhibitors and evaluate results against experimental data reported by Wang et al.

Figure 1: Development of pyridine-based BRD4 bromodomain inhibitors reported by Wang et al. Macrocyclization of optimized fragment [2] improved binding affinity (Ki values from TR-FRET binding assay) about 4x. In addition to improved binding, compound [3] shows an increased cellular efficacy and improved pharmacokinetic properties.


Spark replaces a specified moiety in a given starter molecule with fragments that exhibit similar electrostatic and steric properties.5 Spark’s molecular comparisons are based on molecular fields, computed by the XED force field.6–8 A major benefit of Spark is that it retains a larger amount of 3D information in the search for replacement fragments: instead of scoring only the respective replacement fragment in isolation, the whole final molecule is scored, considering any electronic and conformational effects the new scaffold may have on the rest of the molecule. Replacement fragments are selected from Spark’s fragment databases which are derived from commercially available and literature compound collections. After identifying fragments that contain the required number of attachment points and possess a shape / geometry compatible to the starting molecule, the molecules are minimized using the XED force field to remove bad clashes and unfavorable geometries. A final molecular field calculation and similarity optimization is then performed to yield a final score.

The macrocyclization wizard was used to design BRD4 macrocycle inhibitors. The wizard module provides a quick and easy-to-use GUI workflow with dedicated Spark settings tailored for macrocycle design (Figure 2). For this experiment, we used the X-ray structure of BRD4 in complex with the non-macrocyclic compound [2] (pdb 5UEY) as receptor, and compound [2] as the starter molecule (Figure 2). The bioactive conformation of [2] shows that the ethoxyl group of the pyridone scaffold is in close proximity to the 2,4-difluorophenoxy ring (Figure 3). Additionally, the ligand is partially solvent exposed in this region of the binding site, offering sufficient space to accommodate additional linker atoms. Therefore, we chose to link point 1 (OEt moiety) with point 2 (hydrogen atom) at position 6 of the 2,4 difluorophenoxy ring (Figure 3).

Figure 2: Selection of attachment points in the macrocycle wizard. Attachment point 1 is the ethoxy group and attachment point 2 is hydrogen 6 on 2,4 difluorophenoxy ring. The 3D window (right side) visualizes attachment points and confirms correct selection.

Figure 3: Inhibitor [2] in complex with BRD4 (pdb 5UEY). The close proximity between the ethoxy group and the 2,4 difluorophenoxy ring and the partially solvent exposed volume in this area suggests that linking these features may be feasible.

To bias linker design towards existing chemistry, oxygen was specified as attachment atom for attachment point 1. The second attachment point was left unconstrained. Additionally, the receptor structure was used as an excluded volume to guide linker design and the field points of carbonyl and sulfonyl groups forming hydrogen bonds with the receptor were constrained. The experiment was run with the dedicated ‘Ligand Joining / Macrocyclization’ settings against the fragment databases CHEMBL_common, Common, and VeryCommon, containing altogether about 120K fragments. Results were evaluated in Spark and molecular overlays were rendered in Flare9.

Results and discussion

The macrocycle wizard experiment generated 500 rank-ordered macrocycles derived from fragments merged with the starter molecule. The X-ray structure of the reported and structurally validated macrocycle [3] (Figure 4) shows that upon cyclization the molecule retains its key interaction with the BRD4 pocket, however, compound [3] adapts a slightly different conformation than [2], especially for the 2,4-difluorophenoxy ring. Despite this conformational change, the top scoring Spark results contained several macrocycles closely related to [3] (Figure 5), such as compounds with a butanol (rank 11), propan-1,3-diol (rank 9), 4-aminobutan-1-ol (rank2), butan-2-ol (rank 4), or a 3-buten-1-ol linker (rank 17). Compound [3] was found at rank 59 (Figure 5).

Among the top Spark results, linkers with 4 to 6 atoms were particularly enriched (Table 1). This finding is in good agreement with the experimental data reported by Wang et al., who found that short macrocycle linkers with only 3 atoms decrease the binding affinity by about 50x compared to a 5 atom linker.

Figure 4: X-ray structures of BRD4 in complex with [2] (3UEY) and the experimentally validated macrocycle [3] (3UEX).

Figure 5: Spark macrocyclization results. Among the top scoring results, Spark designed several linkers that are either identical or very similar to compound [3]. ECFP4 values are Tanimoto similarities to [3]. Scores were calculated by Spark and used for ranking of results. Attachment points A1 and A2 as specified in the method section.

Four atom linkers were only slightly less potent (factor 3) than the 5 atom linker, whereas 6 atom linkers showed a comparable binding affinity.4

The top ranked Spark results for each linker size are shown in Figure 6. Interestingly, for the shorter linker sizes (e.g., butan-2-ol with 4 linker atoms) Spark predicted fragments that mimic the hydrophobic volume of the ethoxy side chain of [2]. Omission of parts of this hydrophobic volume may be contributing factors in the observed potency loss for 3 atom linker macrocycles. Adding a Me group to the linker to address this feature of [2] (e.g., compound at rank 6 in Figure 6) may, therefore, result in short macrocycles with improved potency against BRD4.

Table 1: Distribution of linker sizes in top10, top25, and top50 results from Spark (500 total).

Top N results 3 linker atoms 4 linker atoms 5 linker atoms 6 linker atoms 7 linker atoms
10 2 6  1  1 0
25  2 16 4 8 0
50  2 32 12 9 0


Figure 6: Spark macrocyclization results. Among the top 10 results Spark designed compounds with linker sizes between 3 to 6 atoms. The top ranking result for each linker size is shown. Attachment points A1 and A2 as specified in the method section.


This case study demonstrates that Spark can successfully design macrocycles that are identical or very similar to reported BRD4 macrocycle inhibitors.4 The distribution of generated linker sizes was in good agreement with experimental SAR data.4

The Spark macrocyclization wizard is a quick and easy-to-use workflow that generates meaningful and diverse design ideas that can guide macrocycle drug discovery.

References and Links

  2. Wagner, V.; Rarey, M; Christ, C.D. Computational Macrocyclization: From de Novo Macrocycle Generation to Binding Affinity Estimation. ChemMedChem 2017, 12 (22), 1866–1872.
  3. Yu, X.; Sun, D. Macrocyclic Drugs and Synthetic Methodologies toward Macrocycles. Molecules 2013.
  4. Wang, L.; McDaniel, K. F.; Kati, W. M. Fragment-Based, Structure-Enabled Discovery of Novel Pyridones and Pyridone Macrocycles as Potent Bromodomain and Extra-Terminal Domain (BET) Family Bromodomain Inhibitors. Med. Chem. 2017, 60 (9), 3828–3850.
  5. Tosco, P.; Mackey, M. Lessons and Successes in the Use of Molecular Fields. In Comprehensive Medicinal Chemistry III; Elsevier, 2017; pp 253–296.
  6. Vinter, J. G. Extended Electron Distributions Applied to the Molecular Mechanics of Some Intermolecular Interactions. J Comput Aided Mol Des 1994, 8 (6), 653–668.
  7. Cheeseright, T.; Mackey, M.; Rose, S.; Vinter, A. Molecular Field Extrema as Descriptors of Biological Activity:  Definition and Validation. Chem. Inf. Model. 2006, 46 (2), 665–676.
  8. Vinter, J. G. Extended Electron Distributions Applied to the Molecular Mechanics of Some Intermolecular Interactions. II. Organic Complexes. J Comput Aided Mol Des 1996, 10 (5), 417–426.

Spark V10.5 release offers improved usability and flexibility, and new science

I am delighted to announce the release of a new version of Spark™, our scaffold hopping and bioisostere replacement tool. The focus of V10.5 is on advanced workflows and improved database management but also includes new science and many new and improved features.

The most interesting new features are presented below, and I encourage you to try this new release for yourself to see them in action.


  • New wizards to support ligand growing and linking, macrocyclization and water replacement experiments
  • Enhanced Spark database update functionality
  • New pharmacophore constraints
  • Enhancements in search algorithm and advanced options.

New Spark wizards

The new Spark wizards will help you set-up advanced bioisostere replacements experiments in a user friendly and scientifically robust manner.

Figure 1. The new Spark project wizard.
The ‘Ligand Growing Experiment’ wizard (Figure 2) can be used to grow a starter molecule into new space, guided by existing ligands mapping a different region of the same active site. This was possible in previous versions of Spark (see case study Using Spark Reagent Databases to Find the Next Move) but the new wizard makes the workflow easier and more accessible.

The ‘Water Replacement’ wizard (Figure 2) can be used to search for a group which will displace a crystallographic water molecule near your ligand. Again the new wizard significantly improves the workflow for this popular Spark experiment that we have detailed previously (see case study Displacing crystallographic water molecules with Spark).

Figure 2. Results of ligand growing and water replacement Spark experiments.
The ‘Join Two Ligands’ (to  find a linker that joins two ligands sitting in the same active site) and the ‘Macrocyclization’ (to cyclize a molecule by joining two atoms with a linker) wizards are new workflows, which we have been testing internally in the last few years. Full case studies for these workflows are in preparation, but you can see an example of result you can get in Figure 3.

Figure 3. Results of joining two ligands and macrocyclization Spark experiments.
In developing the wizards, a number of additional features to support these advanced experiments have been fine-tuned. These include:

  • New ‘Ligand Growing’ and ‘Ligand Joining / Macrocyclization’ calculation methods to support ligand growing, ligand joining, macrocyclization and water replacement experiments
  • A starter molecule sitting within a protein’s active site can now be downloaded directly from the RCSB
  • Hydrogen atoms can now be replaced in the starter molecule
  • New ‘merge’ functionality to merge molecules in the wizards (when appropriate) and in the Manage/Edit References dialog
  • The starter molecule region selector now includes both 2D and 3D display.

These advanced bioisostere replacement experiments work well because of Spark’s product-centric approach. In Spark, result molecules are compared to the starter molecule, and a similarity score is calculated, only when the new molecules have been formed, minimized, and their fields and field points re-created. This approach, combined with the power of the Cresset XED force field, enables Spark to work with a higher level of precision, by avoiding fragment scoring limitations, allowing for neighboring group effects and for the electronic influence of replacing a moiety on the rest of the molecule and vice versa.

Enhanced Spark database update functionality

Spark V10.5 offers considerable improvements to the Spark database update functionality, to make this process more user friendly and efficient.

The Spark search dialog now alerts you when an updated version of the databases is available, before you start a search. Databases which need updating are marked by a green icon (Figure 4): and if the databases are selected for the Spark search, an ‘Update’ link appears at the bottom of the window, taking you directly to the Spark Database Updater. You will find this particularly beneficial if you use the eMolecules reagent databases where our monthly release schedule provides you with the latest availability information.

Figure 4. The Spark search dialog now alerts users when an updated version of the databases is available. If the databases which need updating are selected for the search, a link appears at the bottom of the window which takes you directly to the Spark Database Updater.
The Spark Database Updater has also been improved. The databases are now categorized as ‘Cresset’ and ‘Cresset reagents’, to make it easy to locate those you wish to update (Figure 5). Furthermore, you can now download or update all the displayed databases in one go by clicking the ‘Install or Update Displayed Databases’ button at the bottom of the window.

Figure 5. The Spark Database Updater now shows different categories of Spark databases and includes a button to install or update all the displayed databases in one go.
Finally, a new ‘sparkdbupdate’ binary is available to enable you to update all or selected Spark databases from the command line.

What’s new in Spark searches

A significant number of improvements and new features to the Spark searches have been made in this release.

Field and pharmacophore constraints

Field and pharmacophore constraints can be used to bias the Spark search towards results which fit the known SAR or your expectations, by introducing a penalty which down-scores results that do not satisfy the constraint.

Field constraints enable you to specify that a particular type of field must be present in the Spark result. For example, you may want to a constrain a positive field where you want an interaction but this can be matched by both H-bond donors and other electropositive features such as the aromatic hydrogens of the compound in Figure 6 – right.

Pharmacophore constraints, new in Spark V10.5, ensure that the desired pharmacophore features are matched by an atom of a similar type in the Spark results. In Figure 6, a pharmacophore constraint was introduced to ensure that all Spark results contain a H-bond acceptor.

While field and pharmacophoric constraints are a powerful way of fine tuning Spark results, we recommend that they are using sparingly, as they will be introducing a bias in your experiment. For example, introducing a pharmacophore constraint on the indazole NH of the PDB 4Z3V ligand in Figure 6 – left would not have matched the aromatic hydrogens of the active ligand in Figure 6 – right.

Figure 6. Left: Ligand from PDB: 4Z3V with pharmacophore and field constraints. Right: Active BTK ligand which satisfies both constraints.
Read more about field and pharmacophore constraints in the Forge V10.5 and Blaze V10.3 release announcements.

Enhancements to the Spark search algorithm

Spark V10.5 also includes enhancements to the search algorithm and associated advanced options, for example:

  • New functionality to weigh specific fields independently when scoring
  • New similarity metrics to provide alternate scoring methods for the alignments
  • New widget for adding field and pharmacophore constraints
  • New ‘Flexibility’ filter which can be applied (together with SlogP and TPSA filters) to the whole molecule when performing a Spark search, to limit the results to the desired physico-chemical space.

Other new features and improvements

This Spark V10.5 release also includes a variety of additional new functionalities and improvements to the interface (Figure 7). These include:

  • New ‘Send to Flare™’ functionality to send either all results, favorite results or selected results to Flare, including as appropriate the starter and reference molecule(s) and the protein
  • New Storyboard window, to capture scenes recording all details from the 3D window that can be easily recalled when needed, including capability to annotate and rename scenes
  • New stereo view functionality
  • New support for touch screen displays and HiDPI
  • New Flexibility column in Spark Results tables
  • Improved performance of Spark database generation
  • Improved ‘View Parent Structure for Selected Result’ functionality now including both substructure and identity search
  • Improved ‘Grid’ button functionality
  • Improved display of protein ribbons, offering a choice of different ribbon styles and capability to show ribbons for the active site only
  • Improved look and feel of the GUI with re-designed toolbars and updated and clearer icons for a more modern and sleek interface.

Figure 7. The Spark V10.5 GUI.

Try Spark V10.5

This release represents a significant improvement in the usability and flexibility of the leading bioisostere application. I encourage you to upgrade your version of Spark at your earliest convenience, and to download the keyboard shortcut guides for Spark V10.5 and Spark V10.5 Molecule Editor.

If you are not currently a Spark customer, please request a free evaluation.

Contact us if you have queries relating to this release.