Inverse Molecular Design & Inference: building a Molecular Foundry

[button url=”” target=”_self” color=”primary”]Back to Calendar[/button]

If you are interested in attending this event, please visit the CECAM website here.

Workshop Description

The overarching theme of this proposed E-CAM Transverse Extended Software Development Workshop is the design and control of molecular machines including sensors, enzymes, therapeutics, and transporters built as fusion proteins or nanocarrier-protein complexes, and in particular, the software development and interfacing that this entails. Several immuno-diagnostic companies and molecular biology experimental groups have expressed a strong interest in the projects at the core of this proposal. The proposed ESDW is transverse as it entails the use of methodologies from two E-CAM Scientific Workpackages: WP1 (Advanced MD/rare-events methods) and WP4 (Mesoscale/Multiscale simulation).

Fusion proteins are sets of two or more protein modules linked together where the underlying genetic codes of each module and the fusion protein itself are known or can be easily inferred. The fusion protein typically retains the functions of its components, and in some cases gains additional functions. They occur in nature, but also can be made artificially using genetic engineering and biotechnology- and used for a wide variety of settings ranging from unimolecular FRET sensors, novel immuno-based cancer drugs, enzymes [1,2] and energy conversion (for example efficient generation of alcohol from cellulose) [3,4]. Fusion proteins can be expressed using genetic engineering in cell lines, and purified for in-vitro use using biotechnology. Much of the design work is focused on how different modules are optimally linked or fused together via suitable peptides, rather than on internal changes of modules. Optimizing such designs experimentally can be done through for example random mutations, but a more controlled approach based on underlying molecular mechanisms is desirable, for which a pragmatic multiscale approach is ideally suited combining bioinformatics and homology, coarse-graining, detailed MD and rare-event based methods, and machine learning. The figure on the front of this proposal is a representative example of a fusion protein sensor designed to bind to a specific RNA nucleic acid sub-sequence, which causes an optimized hinge-like protein to close and in the process bring two fluorescence proteins together allowing the binding event to be observed optically through FRET microscopy.

Nanocarriers (NC) are promising tools for cancer immunotherapy and other diagnostic and therapeutic applications. NCs can be decorated on their surface with molecules that facilitate target-specific antigen delivery to certain antigen-presenting cell types or tumor cells. However, the target cell-specific uptake of nano-vaccines is highly dependent on the modifications of the NC itself. One of these is the formation of a protein corona [5] around NC after in vivo administration. Appropriate targeting of NC can be affected by unintended interactions of the NC surface with components of blood plasma and/or with cell surface structures that are unrelated to the specific targeting structure. The protein corona around NC may affect their organ-specific or cell type-specific trafficking as well as endocytosis and/or functional properties of the NC. Most importantly, the protein corona has been shown to interfere with targeting moieties used to induce receptor-mediated uptake of the NC, both inhibiting and enhancing internalization by specific cell types [5]. Moreover, the protein corona is taken up by the target cell, which may alter their function. Therefore, tailoring the surface properties of the NC to facilitate the adsorption of specific proteins and control the structure of the corona can help to significantly improve their performance. Modification of surface properties, e.g. via grafting olygomers, is also known to affect the preferred orientation of adsorbed proteins and, therefore, their functionality [6]. The molecular design would include the selection of appropriate NC coating and the type of antibody to optimize the NC uptake.

Mesoscale simulation is required to understand the thermodynamics and kinetics of protein adsorption on the NCs with engineered surfaces [7] and to achieve the desired structure with preferred adsorption of the selected antigen. However, the aforementioned issues often require biological and chemical accuracy that typical mesoscale models cannot achieve unless buttressed by accurate simulations at an atomistic/molecular level, rare-event methods and machine learning.

A pragmatic approach towards the enhancement of fusion proteins and NC’s is as follows.

(i) Molecular designs are initially developed and optimized as simple CG models and include the use of information theory and machine learning.

(ii) The solution of the inverse problem of building the fusion protein or the NC-protein complex to match the design requires a multiscale approach combining mesoscale modeling, molecular dynamics, rare-event methods, machine learning, homology, mutation, solvent conditions.

(iii) Iterate steps (i) and (ii) to optimize the design, and in the process collect data for machine learning driven design.

(iv) Final validation using detailed MD, rare-event methods and HPC

The ESDW we plan will over the course of two 5 day meetings with several intervening months produce multiple software modules including the following.
(a) C/C++/Modern Fortran or python based codes to build and optimize simple CG models of fusion proteins or NC-protein complexes using information theory and machine learning.

(b) Semi-automated pipelines to solve the inverse problem of building the fusion protein or the NC to match the design. This will involve interfacing with md/ mesoscale engines such as LAMMPS, Gromacs, OPENMM, EXpresso, rare-event based methods such as PLUMED, and bioinformatics code such as I-TASSER, INTFOLD.

(c) Particle insertion/deletion methods for alchemistry – mutation of amino acids, changes in the solvent and associated changes in free energy properties.

(d) Codes to add corrections to coarse-grained models (bead models/martini) using detailed atomistic data (e.g. potential of mean force for key order parameters, structure factors etc) or experimental data where available.

While this is an ambitious plan, it is worth pointing out that a similar integrated approach to protein development was already made by the lab of John Chodera [8]. While it did not include the focus on fusion proteins or NC-protein complexes or incorporate systematically coarse-graining, it demonstrates both the feasibility of what we propose here and how to achieve practical solutions. Other ideas of a systematic approach to molecular design using MD simulation have been also proposed recently [9,10].



[1] H. Yang et al, The promises and challenges of fusion constructs in protein biochemistry and enzymology, Appl Microbiol Biotechnol (2016)
[2] Bochicchio, Anna et al, Designing the Sniper: Improving Targeted Human Cytolytic Fusion Proteins for Anti-Cancer Therapy via Molecular Simulation, Biomedicines, 5(1),9 (2017)
[3] Y. Fujita et al, Direct and Efficient Production of Ethanol from Cellulosic Material with a Yeast Strain Displaying Cellulolytic Enzymes, Appl Environ Microbiol. 68(10): 5136–5141 (2002)
[4] M. Gunnoo et al, Nanoscale Engineering of Designer Cellulosomes, dv Mater. 28(27):5619-4 (2016)
[5] M. Bros et al. The Protein Corona as a Confounding Variable of Nanoparticle-Mediated Targeted Vaccine Delivery, Front. Immunol. 9, 1760 (2018).
[6] I. Lieberwirth et al. The Role of the Protein Corona in the Uptake Process of Nanoparticles, 24, Supplement S1, Proceedings of Microscopy & Microanalysis (2018)
[7] H Lopez et al. Multiscale Modelling of Bionano Interface, Adv. Exp. Med. Biol. 947, 173-206 (2017)
[8] DL. Parton et al Ensembler: Enabling High-Throughput Molecular Simulations at the Superfamily Scale. PLoS Comput Biol 12(6): e1004728, (2016)
[9] PV. Komarov et al. A new concept for molecular engineering of artificial enzymes: a multiscale simulation, Soft Matter 12, 689-704 (2016)
[10] BA. Thurston et al. Machine learning and molecular design of self-assembling -conjugated oligopeptides, Mol. Sim. 44, 930-945 (2018)
[11] D. Carroll. Genome Engineering with Targetable Nucleases, Annu. Rev. Biochem. 83:409–39 (2014)


Integrating LAMMPS with OpenPathSampling

This module shows how LAMMPS can be used as Molecular Dynamic (MD) engine in OpenPathSampling (OPS) and it also provide a benchmark for the impact of OPS overhead over the MD engine.

Practical application and exploitation of the code

OpenPathSampling uses OpenMM as default engine for calculating the sampled trajectories. Other engines as GROMACS and LAMMPS can be used (despite not yet available in the official release) allowing to exploit different computer architectures like hybrid CPU-GPU and to simulate more complex problems.

In this module we present the source code for the integration of OPS with LAMMPS as well as a benchmark for of a simple test case to show the impact on the performance due to OPS overhead.

Software documentation and link to the source code can be found in our E-CAM software Library here.


ESDW: Topics in Classical MD

[button url=”” target=”_self” color=”primary”]Back to Calendar[/button]

If you are interested in attending this event, please visit the CECAM website here.

Workshop Description

Classical molecular dynamics (MD) is a broad field, with many domains of expertise. Those specialist domains include topics like transition path sampling (which harvests many examples of a process in order to study it at a statistical level [1]), metadynamics (which runs a trajectory with modified dynamics that enhance sampling, and from which free energy profiles can be constructed [2]), as well as various topics focused on the underlying dynamics, either by providing better representations of the interactions between atoms (e.g., force fields [3] or neural network potentials [4]) or by changing the way the dynamics are performed (e.g., integrators [5]).

Frequently, experts in one domain are not experienced with the software of other domains. This workshop aims to combine both depth, by extending domain-specific software, and breadth, by providing participants an opportunity to learn about software from other domains. As an extended software development workshop (ESDW), a key component of the workshop will be the development of modules that extend existing software packages. Ideally, some modules may connect multiple domain-specific packages.

Topics at this workshop will include using and extending modern MD software in the domains of:

* advanced path sampling methods (and the software package OpenPathSampling)
* metadynamics and the calculation of collective variables (and the software package PLUMED)
* machine learning for molecular dynamics simulatons (including local structure recognition and representation of potential energy surfaces)

In addition, this workshop will feature an emphasis on performance testing and benchmarking software, with particular focus on high performance computing. This subject is relevant to all specialist domains.

By combining introductions to software from different specialist fields with an opportunity to extend domain-specific software, this workshop is intended to provide opportunities for cross-pollination between domains that often develop independently.


[1] Bolhuis, P.G. and Dellago, C. Trajectory‐Based Rare Event Simulations. Reviews in Computational Chemistry, 27, p. 111 (2010).
[2] A. Laio and F.L. Gervasio. Rep. Prog. Phys. 71, 126601 (2008).
[3] J.A. Maier, C. Martinez, K. Kasavajhala, L. Wickstrom, K.E. Hauser, and C. Simmerling. J. Chem. Theory. Comput. 11, 3696 (2015).
[4] T. Morawietz, A. Singraber, C. Dellago, and J. Behler. Proc. Natl. Acad. Sci USA, 113, 8368 (2016).
[5] B. Leimkuhler and C. Matthews. Appl. Math. Res. Express, 2013, 34 (2013).


E-CAM related work labeled as “Excellent Science” by the EC Innovation Radar Initiative

The Innovation Radar aims to identify high-potential innovations and innovators. It is an important source of actionable intelligence on innovations emerging from research and innovation projects funded through European Union programmes.

E-CAM is associated to the following Innovations (Innovation topic: excellence science):

    1. Improved Simulation Software Packages for Molecular Dynamics (see link)
    2. Improved software modules for Meso– and multi–scale modelling (see link)

Related to the work of our E-CAM funded Postdoctoral researchers supervised by scientists in the team, working on:

  • Development of the OpenPathSampling package to study rare events  (Universiteit van Amsterdam). Link1
  • Implementation of GPU version of DL_MESO_DPD (Hartree Centre (STFC)). Link
  • Development of polarizable mesoscale model for DL_MESO_DPD (Hartree Centre (STFC)). Link
  • Development of the GC-AdResS scheme (Freie Universitaet Berlin). Link

  • Implementation of hierarchical strategy on ESPResSO++ (Max Plank Institute for Polymer Research, Mainz). Link

Scientific Report from State-of-the-Art Workshop “Large Scale activated event simulations” is available on our website

The workshop scientific report from the E-CAM State-of-the-Art Workshop Large scale activated event simulations that took place on the 1-3 October 2018 in the CECAM-AT Node (Austria), is now available for consultation and download on our website under this link.

Short Description:

The State-of-the-Art workshop in the E-CAM classical molecular simulation work-package (WP1)  brought together 40 participants including scientists from non-academic research centres, to discuss computational approaches capable of addressing time scale problems in complex systems in materials science and biophysics. Scientific discussions at the workshop centred around three fundamental computational challenges closely related to the time scale problem of classical MD simulation: (1)  The calculation of the populations of metastable states of an equilibrium system; (2) The sampling of transition pathways between long-lived (meta)stable states and the calculation of reaction rate constants; and (3) The extraction of useful mechanistic information from the simulation data and the construction of low-dimensional models that capture the essential features of the process under study. The main outcomes from each discussion are described in the workshop report.

Two open discussion sessions revolved on efficient path sampling methods and the identification of reaction coordinates; and how machine learning approaches can be used to make progress in this area. Another important goal of the workshop was to debate about how to facilitate the use of simulation and modelling in industrial settings, with the workshop participants with industrial experience emphasing the importance of detailed project management and, in particular, the need to have very clear agreements about intellectual property rights.

Other scientific reports from State of the Art and Scoping workshops can be hound here:


Topics in Classical MD – Extended Software Development Workshop

E-CAM is organising an Extended Software Development Workshop in Topics in Classical MD from 3 to 12 April 2019, which is a major coding initiative that will combine lectures; coding sessions and hands-on training.

Topics at this workshop will include using and extending modern MD software in the domains of:

  • advanced path sampling methods (and the software package OpenPathSampling)
  • metadynamics and the calculation of collective variables (and the software package PLUMED)
  • machine learning for molecular dynamics simulatons (including local structure recognition and representation of potential energy surfaces).

In addition, this workshop will feature an emphasis on performance testing and benchmarking software, with particular focus on high performance computing.

This is a great opportunity to bring your software development project in all specialist domains of Classical MD and spend two weeks in the beautiful city of Lyon with other peers and with experienced coders. More information and apply through the CECAM website at:


Two papers introducing OpenPathSampling, a software package to study rare events

Two papers introducing to OpenPathSampling (OPS) were recently published : 

Continue reading…


New publication is out: “Unimolecular FRET Sensors: Simple Linker Designs and Properties”


A new publication by researchers at University College Dublin and Boston University including E-CAM funded corresponding author Dr. Donal MacKernan was published in Nano Communication Networks. The measurement of biomarkers and ligands are increasingly used to study transport, signaling, and communication in cells, and as diagnostics/prognostics of disease, or the presence of pathogens, allergens and pollutants in foods, and the environment. Accurate measurement in assays or cellular environments is important, and protein-based biosensors can be used in this context. Using simple Coarse-Grained models of unimolecular fusion protein based FRET sensors of target ligands, the authors address four main questions. Can simple CG models reproduce qualitatively experimental results? Is there an advantage in replacing flexible protein linkers with hinge-like peptides? To enhance the precision of measurement, is it  better to increase or decrease the Föster radius of fluorescent proteins? Is precision enhanced or reduced if the binding energy of the ligand and sensor domains is attractive or repulsive in the absence of the target ligand? The answers are disclosed in the paper.

The publication post-print version is open access and can be downloaded directly from the Zenodo repository here. The publisher’s version can be found at

This work is a particular part of  E-CAM Pilot Project on Food and Pharmaceutical Proteins focused on the development of protein-based sensors and therapeutics. The software used in the publication will soon be published in the E-CAM software library. Industry partners of the Pilot Project include Kerry GroupAPC and others.


Title: Unimolecular FRET Sensors: Simple Linker Designs and Properties

Authors: Shourjya Sanyal, David F. Coker, Donal MacKernan

Abstract: Protein activation and deactivation is central to a variety of biological mechanisms, including cellular signaling and transport. Unimolecular fluorescent resonance energy transfer (FRET) probes are a class of fusion protein sensors that allow biologists to visualize using an optical microscope whether specific proteins are activated due to the presence nearby of small drug-like signaling molecules, ligands or analytes. Often such probes comprise a donor fluorescent protein attached to a ligand binding domain, a sensor or reporter domain attached to the acceptor fluorescent protein, with these ligand binding and sensor domains connected by a protein linker. Various choices of linker type are possible ranging from highly flexible proteins to hinge-like proteins. It is also possible to select donor and acceptor pairs according to their corresponding Föster radius, or even to mutate binding and sensor domains so as to change their binding energy in the activated or inactivated states. The focus of the present work is the exploration through simulation of the impact of such choices on sensor performance.


PLUMED wrapper for OpenPathSampling


PLUMED is a widely used and versatile rare-event sampling and analysis code that can be used with various Molecular Dynamics (MD) engines. It has a very intuitive and versatile syntax for the definition of Collective Variables (CVs), and a wide variety of sampling methods, which accounts for its widespread use. The present module allows PLUMED and OPS to be used together. More details on the module can be found here.

Practical application and exploitation of the code

Transition path sampling simulations and analysis rely on accurate state definitions. Such states are typically defined as volumes in a Collective Variables space. OPS already supports a number of CVs, including the ones defined in the MDTraj python library. PLUMED offers a wide variety of extra CVs, which are enabled in OPS by this module. Many of PLUMED’s dozens of CVs have a biomolecular focus, but they are also general enough for other applications. PLUMED’s popularity (over 500 citations in 4 years after the release of PLUMED2 [1]) is greatly based on the fact that it works with many MD codes. OPS is now added to that list. The PLUMED code is well-maintained and documented for both users and developers. Several tutorials and a mailing list are available to address FAQs. More information about PLUMED is available here.


[1] G. Tribello, M. Bonomi, D. Branduardi, C. Camilloni, G. Bussi PLUMED 2: New feathers for an old bird Comput. Phys. Commun. 185 (2014) 604




The Curse of Dimensionality in Data-Intensive Modeling in Medicine, Biology, and Diagnostics

With Prof. Tim Conrad (TC), Free University of Berlin, and Dr. Donal Mackernan (DM), University College Dublin.


Until recently the idea that methods rooted in statistical physics could be used to elucidate phenomena and underlying mechanisms in biology and medicine was widely considered to be a distant dream.  Elements of that dream are beginning to be realized, aided very considerably by machine learning and advances in measurement, exemplified by the development of large-scale biomedical data analysis for next-generation diagnostics. In this E-CAM interview of Tim Conrad,  the growing importance of diagnostics in medicine and biology is discussed. One difficulty faced by such developments and shared with particle-based simulation is the “curse of dimensionality”. It is manifest in problems such as: (a) the use of a very large number of order parameters when trying to identify reaction mechanisms, nucleation pathways, metastable states, reaction rates; polymorph recognition (b) machine learning  applied to electronic structure  problems – such as neural network based potentials need very high dimensional basis sets; (c) systematic coarse-graining would ideally start with a very high dimensional space and systematically reduce the dimension.  The opportunities and challenges for scientists engaging with industry are also discussed. Tim Conrad is Professor of “Medical Bioinformatics” at the Institute of Mathematics of the Free University of Berlin and head of MedLab, one of the four laboratories of the Modal research campus. MODAL is a public-private partnership project which conducts mathematical research on data-intensive modeling, simulation, and optimization of complex processes in the fields of energy, health, mobility, and communication.  Tim Conrad is also the founder of three successful start-up companies.

In this E-CAM interview with Prof. Tim Conrad, the growing importance of diagnostics in medicine and biology is discussed, including concepts rooted in signal analysis relevant to systematic dimensional reduction, and pattern recognition, and the possibilities of their application to systematic coarse-graining. The opportunities and challenges for scientists of start-up companies are also discussed based on experience.


Continue reading…