Forge V10.3 released

We are delighted to announce the release of a new version of Forge, our computational chemistry workbench for ligand-based design. This version, V10.3, includes over 70 new or improved features.

The development of our applications is always guided by our customers and this release is bursting with features that you have asked for. A few of these are described below but there is no substitute for using the software for yourself to see the new features in action.

Release highlights

  • Support for multiple activities in Activity Miner – find and decipher selectivity as well as activity cliffs
  • Automatic enumeration of unspecified chiral centers
  • Improved conformation hunt settings that gives significantly fewer, lower energy conformations.
  • New integrated FieldTemplater module – find a detailed pharmacophore from diverse ligands
  • Integration with Blaze for virtual screening – submit molecules to Blaze and retrieve results straight into Forge
  • New radial plots enable rapid assessment of compounds properties against project or corporate profiles

Activity Miner

Activity Miner has been a huge hit since its release last year. You told us that you enjoy the powerful yet intuitive interface, that it has helped you to find and understand critical regions in your SAR and that it has helped your medicinal chemists rapidly assess the SAR landscape of projects. However, the number one request has been for selectivity cliffs – the use of multiple activities to look for changes that disproportionately change one activity relative to another. In this release of Forge we are have included this functionality.

Multiple activities

Activity miner now supports multiple activities and displays activity cliff information for all activities and all compounds in a matrix of all compounds or relative to a single compound. All the new data created a major challenge – how were we going to present multiple activity cliffs using multiple similarity metrics for many compounds in an easy to understand way? However, we think we have succeeded in the task that you set us (below) but please tell us how well it works for you on your data!

Forge_10.3_release_am_pic

Use 2 activities or even more!

There is more new science in this release beyond selectivity cliffs. In particular we’ve been looking at our treatment of molecules with an unspecified chiral center and how we perform conformation hunting. In this release we now enumerate all the possibilities in molecules loaded with unspecified chiral centers. Like selectivity cliffs, this feature has been a high priority for our users and we are delighted to introduce this new functionality.

Conformation hunting

Conformation hunting is critical to our alignment methods and hence always something that we are looking to improve, even though independent tests show that our performance is one of the best. Our research has shown that we can reduce the number and increase the quality of conformations we use with a couple of small changes to our settings. By improving the minimization of conformations, combined with taking a smaller energy window for accessible conformations, you get significantly better populations in most cases. However, these changes come with a penalty – the time taken for this type of conformation hunt is increased. As a result we’ve changed how our default ‘Accurate’ methods work to use the new settings but retained the previous settings for ‘Quick’ and ‘Normal’. The great thing is that with our standard calculation dialog box you can choose your own settings and have these quickly available for later use.

Forge_select results quality desired

User interface

The new user interface includes custom plots, radial plots of properties, a manager for activity and the associated error data, ribbon view of proteins and auto-color of selected molecules which are requests from customers. However, one new feature that we wanted for use on our internal work, and especially our consulting projects, is a simple script interface to changing some of the values in the molecule table.

Forge_10.3_release_3D_pic

Column script editor

The column script editor provides a JavasScript interface to molecule and table data and enables modification, formatting or creation of values in the table. A simple example might be to remove the ‘>’ symbol from activity values before using them in a 3D-QSAR experiment, or coloring all values in a column called ‘HERG_IC50’ that are below 10 as red. Once developed (and we are happy to help you get your script working quickly), these scripts can be set to run on every project or exported for use by colleagues running Forge or Torch.

Forge_10.3_release_script_editor

This release is a major advance for Forge so we encourage you to upgrade at your earliest convenience.

If you are not currently a Forge customer then now is a great time to try a free demo.

Contact us if you have queries relating to this release.

Kick-starting stalled projects

When a project runs into patent or toxicology issues, scaffold hopping can be a fast track to new chemical space

In the search for a compound that is efficacious, well tolerated and novel most drug discovery projects hit a hurdle or two. In most cases the problem is not in the activity of the lead compound, but in its physicochemical, ADMET or intellectual properties. Sometimes this can be ameliorated by small changes to the structure, but in most cases this is not sufficient. In this case the best solution is usually to move the project into new areas of chemistry towards compounds that have a better ADMET profile, better IP or better properties. The trick is to do this without losing all of the hard-won knowledge on SAR around the existing series and especially not losing the hard-won activity of the lead.

A common approach is to ‘scaffold-hop’ – change the ‘core’ or ‘scaffold’ of an active molecule while retaining the attached R groups as much as possible. In the ideal case the resulting molecule will have a biological activity very similar to the original, but its different core leads to a new ADMET and IP profile, solving the problems with the original compound. Since Cresset’s field similarity approach describes the binding preferences of a molecule, what we want to find is an alternative core that preserves the field pattern of the original.

Cresset consultants have extensive experience in applying this technique to help clients to kick-start a new project or ‘un-stick’ an existing one. We use our synthetic and computational experience together with our software to analyze your data and suggest new chemical series that will take you in the direction that you want to go.

We may start by developing 3D templates or pharmacophores, or developing models of your SAR using Activity Miner or FieldTemplater. We would then go on to use Spark to suggest novel core replacements that we refine in Forge. Alternatively, we may work closely with your synthetic chemists to define all the possible scaffolds that can be introduced into your molecules. We then use this information in Spark, Forge or BlazeGPU to find the best possible option. Either way, we get your project moving in the right direction, delivering the best possible combination of chemical and biological properties. Throughout this process we work closely with you to ensure that our suggestions fit your project goals and synthetic capabilities.

In the pharmaceutical industry, confidentiality is of fundamental importance to our clients. For this reason, our most impressive work is usually not publishable, so we carry out retrospective analyzes of published projects in order to provide examples of our capabilities.

In spring 2013, Tim Cheeseright, Cresset’s Director of Products, presented a detailed study on the use of Spark for scaffold hopping from the core of Sildenafil, in a comparison with a published technique from Pfizer. He showed that Spark not only found all of the scaffolds identified by the Pfizer team, but located additional known active cores while only using a fraction of the computation time.

At this year’s Cresset North American user group meeting the independent medicinal chemistry consultant Alfred Ajami gave a presentation describing the use of Spark to retrospectively analysze two case studies from the bioisosteric replacement literature and to evaluate its performance in a number of projects that he has been involved with. You can see the full presentation ‘Pharmaceutical hunts with Spark: case studies from the literature and current campaigns to develop immunokinase inhibitors’, Alfred M Ajami, DCAM Pharma Inc.

These studies show the power of the Cresset methods to free your drug discovery program from being stuck in a dead-end series. contact us for more information about how Cresset Consultants can help to kick-start or un-stick your project.

What’s new with the Cresset product line up?

We have made some major changes to the names of our products recently and are thrilled to introduce a new suite of Next Generation Chemistry Software. The suite encompasses five products: Forge, Torch, TorchLite, Blaze and Spark. This new product line up is designed to fit more closely with your computational and medicinal chemistry needs and we encourage you to take a look and let us know what you think.

Some of the products have simply been re-launched with fire-themed names like Blaze and Spark, but others will show significant changes over previous products and now include exciting new features.  If you’re already using Cresset software or you’ve grown accustomed to the old product line, you might want to take a look at the product guide below to see how things have changed. We’re launching the new products together but it will take a few weeks for the developers to finalize all the releases so please bear with us during the change over.

Old Product Name New Product Name Description and Features
FieldTemplater and FieldAlign together Forge Forge is a powerful computational suite to understand SAR and design. It brings together all the functionality of FieldAlign and FieldTemplater together with new science and new capabilities. It helps you:

  • Decipher complex SAR and communicate the results.
  • Design better molecules based on predictions you can trust.
  • Prepare detailed pharmacophores.
  • Virtually screen 10 000 compounds on your desktop.
  • Generate ADME and off target activity profiles.
FieldAlign Torch Torch is a powerful design and 3D SAR tool for medicinal chemists. Torch represents a new opportunity for our medicinal chemistry customers to use a product that is dedicated to the tasks that you want to do.  It helps you:

  • Perfect the design of new lead compounds, exploring a range of lead optimization ideas.
  • Get the most from your lab time by prioritizing compounds for synthesis.
  • Design focused libraries for synthesis or initial screening.
  • View ADME profiles and off-target activity prediction on all designs.
  • Use powerful predictive QSAR models from Cresset’s Forge.
FieldView TorchLite  TorchLite is a free 3D molecule viewing, editing and drawing tool.  It helps you:

  • Sketch your chemical series to study how field and activity patterns are related.
  • Import and compare up to 10 000 compounds from SDF, MOL2 or Cresset applications.
  • Convert from 2D to a minimized 3D conformation.
  • Clone and compare molecules side by side or overlaid using Cresset’s unique field technology.
  • View virtual screening results.
FieldScreen Blaze  Blaze is an amazing ligand based virtual screening tool.  It helps you:

  • Increase the diversity of your project’s leads and backups.
  • Jump into new areas of chemical space.
  • Improve the lead-like properties of hits.
  • Virtually screen 10 million structures in a few hours.
  • Design diverse libraries of compounds for synthesis or biological screening.
FieldStere Spark Spark is an exciting and powerful way of generating novel and diverse structures for your project. It helps you:

  • Generate highly innovative ideas for lead molecules in new areas of chemical space, overcoming the ‘chemotype trap’.
  • Use filters to find the results with the right mix of physicochemical properties and biological activity.
  • Tailor results by selecting the chemistry allowed for the replacement moieties.
  • Visualize results in detail side-by-side, or cluster similar chemical scaffolds.
  • Search for moieties from real, published and unexplored compound databases or create your own custom databases.
Xed Tools Xed Tools Nothing’s changed here!  The XED molecular mechanics force field is still at the heart of Cresset’s field technology and you can still use Xed Tools for conformation generation and ligand minimization.

Still have more questions?  Take a look at the FAQs below, or contact us to learn more about the product line changes and how they work for you.

I already use some of the old products.  What happens to my yearly license?

Nothing changes for now.  You can continue to use your current software license(s) without any interruption in service. When your license is up for yearly renewal, we’ll be in touch to see how your needs align with the new products and then you can decide which products work best for you.

Do these new products work with the same operating systems and interfaces as the previous product line up?

Yes, all of new products work exactly the same way as they did previously.  The table below gives you a good idea of which products work with which operating systems and interfaces.

Operating Systems Interfaces
Product Windows Linux Mac
Forge
Torch
TorchLite
Blaze
Spark
Xed Tools

I just tried a demo of one of the old products. Do I need to try a new demo the new products too?

That depends on which product you’re interested in. If you’ve previously tried FieldTemplater or used FieldAlign and wanted to do more then you will definitely want to check out a new demo of Forge because we’ve added lots of cool new features that were designed specifically with computational chemists in mind. If you’ve tried a demo of FieldStere, FieldScreen, XedTools, or any of the interface options you probably don’t need a new demo right now because all we’ve changed is the product name.   Feel free to contact us if you have questions about your product demo.

Why did you rename the products with fire-themed names?

We’ve been planning to align our product line up more closely with your needs for a long time, and with the launch of Forge, we decided it was also a good time to update a few other things.  Plus we love the fact that our fiery Cresset logo now matches our new product names!

Some of the links on your site don’t work anymore.  What’s happening?

Sorry, this is our mistake and we apologize for the inconvenience. We’re doing our best to ensure that our website gets updated right along with the product changes but you’ve obviously found a mistake.  If you find a broken link, or other problem, please let us know and we’ll fix it right away.

 

Forge Sneak Peek

We’ve been working steadily over the last six months to improve on our offering to Computational Chemists and other power users. The result is Forge, a new application that combines all the features of FieldAlign and FieldTemplater into a single new package and adds a heap more of the functionality you’ve asked for. We are really pleased with the result and as we near release we thought we would let you in on the secrets that are coming soon.

The primary role of Forge is to help with decoding structure activity relationships and generating new molecule designs with meaningful activity predictions wherever possible. On the design side we’ve revamped the molecular editor to give you the flexibility to design molecules using fields in an interactive environment. Take a look at the screenshot below:

Forge Molecular Editor

The screenshot shows the new drawing buttons on the left with ring templates which will please all those that asked for them and make drawing far less painful. There is a new button on the right “Align” will align the molecule that is present in the editor to the reference molecule giving you immediate feedback on a new design and giving you the ability to manually edit alignments. We call this “score while you draw”!

To enhance the understanding of structure activity relationships in your molecules we have introduced automated analysis of field patterns using a QSAR method and new alignment methods to give you more ligand centric views as opposed to the protein centric view that we have traditionally generated. The QSAR that we are introducing is a completely new 3D method that uses the field points around your ligands as sampling points for the field of a ligand. In effect we generate a irregular, non-lattice grid around the ligands which is then used to generate field samples on every molecule in the training set. Using PLS acrosss this dataset gives an indication of which regions of the field are contributing to the observed changes in activity but also generates a way of predicting activities for compounds of unknown activity. The QSAR model that you create can be exported to be used by others or used to help the design process including in the “score while you draw” feature. The QSAR is completely integrated into Forge with all the statistics that you would expect from a method of this type (see the screenshot below for an example). Of course 3D QSAR is not a panacea but in our hands this methods has greatly enhanced many of the customer projects that we have worked on.

Forge screen shot showing model information

Underlying these functionality improvements is a new version of our XED molecular mechanics force field that introduces some fundamental changes as well as field pattern improvements. The biggest change in the force field is the introduction of a completely analogue nitrogen atom type that can transition from planar to pyramidal depending on the degree of hydbridisation with surounding atoms.

Lastly, Forge is firmly a Cresset application and comes with the intuitive user interface that you have come to expect from us, plus  a comprehensive wizard to guide you through the process of setting up experiments.

Forge showing a FieldAlign type of interface
Want more information about Forge?  You can sign up for our newsletter to receive product announcements or contact us to learn more.

FieldTemplater v3.0 Released

The release of FieldTemplater v3.0 completes our two year project to introduce parallelized operation to our desktop applications.   All our applications will now use all the computing resources available on modern desktop machines, giving significant speed increases for all users. FieldTemplater is not limited to just your local machine; using remote “FieldEngines” you can further increase the speed of calculation. Simply specify the location of the remote resource – your neighbors Windows box or Amazon cloud or a local compute cluster- and FieldTemplater will connect to and use these resources to give a massive speed bonus to your calculation.

In addition to a completely new architecture, FieldTemplater V3.0 comes with numerous improvements to the user interface to enhance your experience.   Just like all our other applications, we provide a simple wizard to guide you through the process of setting up and starting your experiment, however, the individual steps have been cleaned up giving fewer mouse clicks and simpler operation. Should you step away from the wizard, you will find we have greatly improved the interaction with other applications through the introduction of extensive copy and paste operations.  You will now have significant control over the conformation space used in calculations by simply copying the conformations of a molecule into FieldView, editing them to remove undesirable conformations, then pasting them back into FieldTemplater.

FieldTemplater’s new preferences section gives control over the look and feel of the application including, for the first time, the option to generate a stereo view in the 3D window. We’ve provided support for the four most common methods of generating stereo images but if we’ve missed one then just let us know.  Just like FieldView and FieldAlign, FieldTemplater enables the creation of high resolution images to enable communication of your results with colleagues or with the wider community.

This release also brings FieldTemplater onto the Mac, expanding our range of native Mac applications. We are now close to our goal of releasing Mac native versions of all of our desktop applications.

Picture of 4 p38 actives in a template

The best template found from 4 known p38 actives (pdb codes 3gcq, 3gpo, 3nmw, 3kji4). The template is remarkably similar to the alignment obtained with protein crystal structures and clearly shows that two compounds do not interact with the hinge region.

If you like what you’re hearing you can try the new version of FieldTemplater for free, just register here and we’ll send you a 1 month license together with download links.  Existing customers should contact their account manager for details.

Cresset and Redx Pharma announce Major Drug Discovery Collaboration

Welwyn Garden City, UK (24th Aug 2011) – Cresset and Redx Pharma are pleased to announce that they have signed a major drug discovery collaboration. The collaboration gives Redx Pharma access to Cresset’s publicly available and proprietary computational chemistry technologies for use on their portfolio of drug discovery programs. In addition Cresset will provide consulting services in the area of candidate selection across a broad range of therapeutic areas and targets.

Rob Scoffin, CEO of Cresset said, “Cresset has developed an extensive range of consulting services around the critical areas of library design, compound design, virtual screening and lead optimization. In this case we will initially be focussing our skills on helping Redx with their lead candidate selection process, whilst the collaboration also gives Redx the ability to access our expertise, in a very flexible manner, across the full range of computational Drug Discovery services.”

“Cresset is a valuable partner in our Drug Discovery programs”, commented Dr Neil Murray, CEO of Redx Pharma, “Their deep knowledge of computational chemistry and its application to drug discovery is enabling us to progress multiple projects across a wide range of target classes very quickly and cost-effectively”.

The Redx programs make use of a number of Cresset’s computational chemistry tools. In particular FieldTemplater and FieldAlign are used respectively to provide binding hypotheses and predictions of likely biological similarity for sets of molecules in various lead series.

About Cresset Group Ltd

Cresset develops software for calculating and comparing the molecular Field characteristics of chemical compounds. Field technology uses the surface properties of molecules to evaluate their activities and properties, rather than relying on 2D structure similarity, which enables Cresset’s users to find more interesting, novel and relevant results than other methods. Cresset’s Fields provide a smarter, structure independent way of hit-finding, lead switching and lead optimization in drug discovery and other chemistry-based research projects. Cresset’s field technologies have been successfully applied to a very wide range of target classes, with and without structural information, on over 100 projects for major pharmaceutical and biotechnology companies.

About Redx Pharma Ltd

Redx Pharma’s Redox Switch approach is being used to generate improved versions of existing drugs with patient benefits including greater efficacy, fewer side effects and better ease of use.  The company’s pre-clinical drug pipeline has lead programmes for indications including cardiovascular, antibiotics, influenza and neuropathic pain.  Redx’s IP portfolio includes additional programmes in cancer, diabetes, immune disease and a range of central nervous system (CNS) disorders.  The Redox Switch technology was invented by Redx Pharma founders Dr. Peter Jackson and Dr. Derek Lindsay. Redx forerunner company Bradford Pharma was established in mid-2005, commenced operations at MerseyBio in mid-2007 and became Redx Pharma in late 2010. See: www.redxpharma.com

Cresset August 2011 Newsletter

European User Group Meeting Sept 21 – 22, 2011

Cresset User Group Meeting

There are a limited number of hotel rooms available, so register now to avoid disappointment. 

More information can be found here…


FieldTemplater V3.0: Coming Soon

The upcoming release of FieldTemplater V3.0 provides an improved user experience and massive speed increases.  Due for release on September 1st, this new version provides all of the accuracy you expect of Cresset software, while using remote calcuation resources to dramatically reduce the time for results.

More information can be found here…


ACS National Meeting Denver

Cresset is looking forward to attending the fall 2011 National Meeting and Expo. You can find us at booth 1817, or attend one of the four presentations Cresset will be giving.  This will be a great opportunity to learn about  the latest in Chemistry research, network with other researchers and watch some high quality presentations.

More information can be found here…


SCI – RSC Med Chem Symposium

Two members of the Cresset team will be attending the SCI-RSC Medicinal Chemistry Symposium this Sept 11-14, 2011.  Come find us at booth 8, where you can register for our Design a Molecule contest, ask questions of our medicinal chemistry experts, or register for a free software demo.

More information can be found here…


Design a Molecule: Round 3!

We’re pleased anounce that our third Design a Molecule Contest will begin August 25, 2011.  The rules are simple: register, receive free access to FieldAlign software for the duration of the contest, design the best molecule and win an iPad2!

More information can be found here…

Sneak Peek – FieldTemplater v3.0

The upcoming release of FieldTemplater provides an improved user experience and massive speed increases.

FieldTemplater will now use all the available cpu cores on your desktop machine giving significant speed increases over previous versions. Running a simple project with 5 small molecules and quick settings now completes in less than 5 minutes on a quad core workstation. A project with 4 Chemokine ligands gives a similar speedup but more interesting is to bring in extra calculation resources.

The new FieldTemplater, in common with FieldAlign and the upcoming release of FieldStere can use remote calculation resources to provide massive speed increases over earlier versions. For example, running the project with 4 Chemokine receptors with 50 remote processes gives results in a few minutes instead of the multiple hours that it used to take. Of course you are not limited to using 50 remote processes, using 100 or more would be faster still.

FieldTemplater v3 running on multiple processors

The increased speed comes with the same high degree of accuracy that you would expect from us, giving you reliable, reasonable hypotheses for how your molecules relate in three dimensions, and predicting bioactive conformations and pharmacophores in the absence of protein crystal data.

We have improved the user interface for this release giving a smoother ride for most users. This will be the first release of our GUI products that supports stereo in the 3D window; we’ve implemented 4 different types (see picture below) that we hope cover most systems but please let us know if we’ve missed one.

Options for Stereo in FieldTemplater 3

A feature that we’re just finalizing is transparent surfaces. Initially impressions suggest these could significantly improve the interpretation of the electrostatics around ligands, particularly when comparing two diverse molecules. Below is a look at the electrostatics around Maraviroc using the new rendering (red is positive, blue is negative regions). Please let us know what you think.

FieldTemplater transparent surface around Maraviroc

FieldTemplater is due for release September 1. If you would like us to keep you up to date on progress then drop us a line: support@cresset-group.com, or follow us on Twitter (@cressetgroup) or join our Linkedin group.

CASE STUDY: SAR Analysis and Lead Generation

Chronic obstructive pulmonary disease (COPD) is a common, progressive disorder of increasing prevalence in industrialized countries. Several drugs have been marketed so far, and several molecular targets have been considered for therapeutic intervention. This project explored the SAR of known PDE4B inhibitors extracted from Thomson Reuters’ Integrity database and used this to derive models that enabled scaffold hopping.

We  first assumed that all molecules act at the same site on the same target. Cresset’s FieldTemplater used the four chosen compounds to find a common Field pattern across 200 representative conformations of each compound.

SAR Analysis 



Template Generation

 

1.Ligand based virtual screening
2.Scaffold fragment replacement

Two separate experiments were performed using the most active 654921 seed structure; firstly a virtual screening experiment to identify bioisosteres from the Cresset databases (containing 4.5M drug and drug-like compounds), and secondly fragment replacement of a section of the 654921 molecule. Some of the results of these experiments are shown below

These results include known active compounds as well as several novel chemotypes, which can then be taken forward as leads in the therapeutic prevention of COPD.

The project reported in this case study was originally presented as a joint poster between Thompson Reuters Scientific and Cresset at the EFMC Conference in Brussels in Sept 2010. Finch, C., Prous, J., Hoffman, R., BUckley, G., Gardner, S., & Vinter, A.

CASE STUDY: Using FieldTemplater in Library Design

There is a delicate balance to be struck in library design between identifying all the chemical scaffolds that are potentially active, whilst retaining a manageable library of compounds that is tractable and cost-effective for routine screening. Field templates can be used in library design to predict activity of compounds both at therapeutic targets as well as at known toxicity targets such as CYP 2D6 and hERG. A range of templates can be derived and used as ‘lenses’ to counterscreen an aggregated library of compounds derived from multiple congeneric series and other sources.

FieldTemplater was used in the example shown to select to select a diverse library of potential H3 antagonists. A series of seven highly active H3 antagonists were identified from the literature and aligned in their bioactive conformations to generate a consensus field template (shown bottom). As confirmation of the predictive capability of this template, the field match score was compared against the known activity (Ki) scores of 68 further H3 antagonists described in the scientific literature and outside the original training set. A good match of fields to activity was confirmed.

The H3 template was then used to counterscreen Cresset’s 4.5M compound collection to identify potential H3 antagonists. A large number of matches were identified, with 68 distinct chemical scaffolds. Since chemical scaffolds that can be expected to show liabilities for serious off-target or toxicity effects should be avoided, the compound matches were also screened against toxicity Field templates for CYP 2D6 and hERG activity. Approximately 4% of the compounds were rejected due to potential 2D6 toxicity and a further 8% due to potential hERG toxicity.


Field based methods can also be used to predict novel bioisosteric compounds that will exhibit the same activity when key fragments of their structure are replaced. Such a tool was used to replace the central core as an alternative library method in order to generate a novel scaffold replacement library. The results of this analysis can be seen in the graph shown below.

The highlighted structures on the graph represent some of the most active known H3 antagonists from the literature and the blue structures represent novel compounds generated by the software. The graph shows a number of novel compounds with diverse central cores that have significantly higher predicted activities at H3 (as shown by higher field similarity score). Five of the more interesting compounds (all of which are novel and have high similarity scores) have been highlighted in yellow.

These highlighted compounds would be ideal candidates for inclusion in the final library as they combine innovation with chemical tractability and high predicted activity. Interestingly, the 2D similarity score of most of the dataset, including all of the highlighted molecules, is less than 0.7, which is a de facto cut-off for 2D based scoring methods. This means that most of these structures would be very unlikely to be considered in a traditional library design process as there would be no reliable way to predict their activity.