International
Conference on Mathematical Multiscale Modeling in Biology
Guanacaste, Costa Rica, October 21-24, 2019
CHAIRS:
Andrzej Kloczkowski,
Ph.D., Nationwide Children’s Hospital & The Ohio State
University
Grzegorz Rempala,
Ph.D., The Ohio State University
Marek Cieplak Ph.D., Institute of Physics Polish Academy of Sciences
COORDINATOR:
NOTE: All times listed are Central Standard Time (local time)
|
Monday,
Oct 21- Plaza Bar (In front of the show stage) |
|
|
6:00pm |
● Registration ● Welcome ● Reception |
Tuesday, Oct 22 [Young Investigators Session I, Chairs: G.
Rempala, M. Pietrzak, R. Jernigan]
|
9:00-9:30am |
Colin Klaus, The
Ohio State University “Multiscale Approaches in
Visual Transduction: A Novel Homogenized
Diffusion Model for Cone Photoreceptors” |
Chair: Rempala |
|
9:30-10:00am |
Maciej
Pietrzak, The Ohio State University “Application
of information-theoretical approaches for the analysis
of human immunome” |
|
|
10:00-10:30am |
|
|
|
10:30-11:00am |
Maksim
Kouza, Nationwide Children’s Hospital “Assessing the Rate of Fibril
Formation using Steered Molecular Dynamics Simulations |
|
|
COFFEE BREAK |
||
|
11:30-12:00pm |
Tamara
Bidone, University of Utah “Multiscale Models of
Integrin-based Mechanosensing” |
Chair: Pietrzak |
|
12:00-12:30pm |
||
|
OPEN DISCUSSIONS/FREE
TIME |
||
|
6:00-6:20pm |
|
Chair: Jernigan
|
|
6:20-6:40pm |
Aiyana
Emigh, University of Califormia- Davis Predicting
Arrhythmogenicity: Structural Modeling of Safe and
Unsafe hERG Blockers |
|
|
6:40-7:00pm |
|
|
|
7:00-7:20pm |
||
|
7:20-7:40pm |
Pranav Khade, Iowa
State University Using
alpha shapes to characterize protein packing and
capture the multiscale aspects of allostery |
|
|
7:40-8:00pm |
Xianli
Jiang, University of Texas- Dallas Unraveling
and designing signal-response connections using direct
couplings from a global coevolutionary model |
Wednesday, Oct 23 [Young Investigators Session II, Regular
Session I,
Chairs: A. Kloczkowski, K. Fidelis]
|
9:00-9:30am |
|
Chair: Kloczkowski |
|
9:30-10:00am |
Grzegorz
Rempala, The Ohio State University Multiscale Stochastic Models of
Transcription/Translation |
|
|
10:00-10:20am |
Komla Gnona,
Natiowide Children's Hospital Developing
genetic biomarkers for polygenic conditions: An
Application to Neonatal Complications in Preterm
Infants |
|
|
10:20-10:40am |
Daniel Kool,
Iowa State University
“Investigating the Changes in Amino Acid Properties in
the Evolutionary and Multi-scale Context” |
|
|
COFFEE
BREAK |
||
|
11:00-11:30pm |
Jayajit Das,
Nationwide Children’s Hospital and The Ohio State
University “Modeling
formation of biofilms by a bacterial pathogen in
vitro and in vivo” |
|
|
11:30-12:00pm |
Leonor Saiz, University of
California-Davis
“Multilevel modeling of cellular networks in
biomedicine” |
|
|
OPEN
DISCUSSIONS/FREE TIME |
||
|
6:00-6:30pm |
Mateusz
Sikora, Max Planck Institute of Biophysics “Truss-like
arrangement of cadherins is responsible for desmosome
strength” |
Chair: Fidelis |
|
6:30-7:00pm |
Sergei Grudinin, Inria/CNRS “Coarse-graining protein dynamics and representation for integrative structural bioinformatics |
|
|
7:00-7:30pm |
Olivier
Lichtarge, Baylor College of Medicine
“The
Skinny From All of Us to Each of Us: A Calculus
of Fitness Suggests Simple Rules for Complex Diseases” |
|
|
7:30-8:00pm |
Marek Cieplak, Institute
of Physics Polish Academy of Sciences “The Networks of the
Inter-basin Traffic and Emergence of Knots in
Intrinsically” |
|
|
8pm GALA DINNER |
Ocotal Steak House |
Thursday, Oct 24 [Regular Session II, Chairs: A. Joachimiak,
M. Cieplak]
|
9:00-9:30am |
|
Chair: Joachimiak |
|
9:30-10:00am |
Krzysztof Fidelis,
University of California- Davis
“Sequence-structure relationship in proteins: An amino
acid structural neighborhood perspective” |
|
|
10:00-10:30am |
William Stewart,
Nationwide Children’s Hospital and The Ohio State
University “Multi-scale
Modeling of Serial Measurements in Survival Data” |
|
|
10:30-11:00am |
Cezary Czaplewski,
University of Gdansk
“Using
coarse-grained UNRES model for simulations of protein
structure and dynamics" |
|
|
COFFEE
BREAK |
||
|
11:15-11:45pm |
Ilya Vakser, University
of Kansas
“Docking of Protein Models” |
Chair: Cieplak |
|
11:45-12:15pm |
Robert Jernigan, Iowa
State University “Using
High Order Coevolution Correlations to Identify Sites
for Compensating Mutations to Rescue Function” |
|
|
12:15-12:45pm |
Andrzej Joachimiak,
University of Chicago “Addressing
Ligand Binding Promiscuity” |
|
|
12:45-1:15pm |
Yinglong Miao, University
of Kansas
“Gaussian Accelerated Molecular Dynamics (GaMD):
Bridging gaps in multiscale simulations” |
|
|
1:15-1:45pm |
Andrzej Kloczkowski,
Nationwide Children’s Hospital and The Ohio State
University “Modeling structure, stability and
dynamics of proteins and protein aggregates” |
|
|
1:45pm |
Closing
Remarks |
|
Abstracts
SPEAKER:
Bidone, Tamara (University of Utah)
TITLE: Multiscale Models of Cell Adhesions Mechanobiology
ABSTRACT: Cell-substrate adhesions are specialized
regions of the plasma membrane that couple the cytoskeleton
network to the extracellular environment. They are critical in
several cell activities, including tissue morphogenesis,
homeostasis, cell migration and wound healing. Central
elements of cell-substrate adhesions are transmembrane receptors, that form physical links
between cell cytoskeleton and external binding partners.
Integrins are transmembrane receptors that sense, resist and
transmit cytoskeletal contractility to the substrate and
respond to substrate rigidity via changes in conformation and
ligand binding affinity. Integrin resists to contractility and
this makes the receptor a cell mechanosensors, responding to
mechanical cues with specific kinetic activities.
Unfortunately, the biophysical and conformational mechanisms
by which adhesions reinforce in response to stress are not
fully understood. In this study, we developed a multiscale
model of adhesions formation based on coupling Molecular
Dynamics, coarse-graining techniques and Brownian Dynamics
approaches in order to study mechanisms of adhesion
reinforcement. Our model shows thaTITLE: integrin
activating mutants extend more easily under tension;
conformational rearrangements in
proximity of the ligand binding domain of integrin underlie
integrin extension; catch-bond kinetics is responsible for
integrin-based mechanosensing, leading to enhanced cell
spreading and traction stress. These results provide important
insights into the biophysical principles and mechanisms of
adhesion mechanosensing, with functional consequences on both
cell and tissue physiology.
SPEAKER: Cieplak, Marek
(Institute of Physics Polish Academy of Sciences)
TITLE:
The
networks of the inter-basin traffic and emergence of knots in intrinsically disordered proteins
ABSTRACT: The equilibrium dynamics of the intrinsically
disordered proteins is thought to consist of transitions
between many basins in the free energy landscape whereas
structured proteins stay in the vicinity of one native basin.
We demonstrate this picture explicitly by studying networks
defined on the discretized plane: conformational end-to-end
distances vs.
radii of gyration. The bin sizes are
defined by time scales that span orders of magnitude.
The networks, derived from all-atom and coarse-grained
molecular dynamics simulations, are nearly scale invariant.
The bin representation also provides insights into the folding
process
of the structured
proteins and identifies regions of hindrance to folding.
Transient
knotted structures are expected to arise during the volatile
evolution of intrinsically disordered eptide chains. We show
that this is indeed the case for sufficiently long
polyglutamine tracts and α-synuclein. The polyglutamine
tracts are fused within huntingtin protein that is associated with the Huntington
neurodegenerative disease. We show that the presence of
knots in the tracts hinders and sometimes even jams
translocation, especially when the knots are deep. The knots in
polyglutamine may form in tracts exceeding about 40
residues. This fact explains the existence of a similarly
sized length treshold above which
there is an experimentally observed toxicity at the
monomeric level. We also discuss emergence of knots in
α-synuclein. We show that these knots are either shallow or
deep and last for about 3 – 5 µs, as inferred from an
all-atom explicit-solvent 20 and 30 µs trajectories. We
discuss conformational biasses that take place in
α-synuclein and contact formation during aggregation of two
chains of this protein. We then discuss several aspects of
dynamics of knotted structured proteins as assessed within a
Go-like model. In particular, we argue that folding under
the nascent conditions is essential to fold to a structure
that is deeply knotted.
In
collaboration with: M. Chwastyk, Ł. Mioduszewski, B. de
Aquino, A. Gomez-Sicilia, M. Carrion-Vazquez, P. Robustelli,
and Y. Zhao.
SPEAKER: Cezary
Czaplewski (University of Gdansk)
TITLE:
Using
coarse-grained UNRES model for simulations of protein
structure and dynamics
ABSTRACT: Computer
modeling is routinely used for the prediction of unknown
protein structures and simulation of protein dynamics.
Coarse-grained models of proteins allow for an extension of
the simulation time- and size-scale by orders of magnitude
compared to all-atom models. The coarse-grained UNRES model is
a highly reduced model of polypeptide chains, with two
interaction sites per residue: united side chains and united
peptide groups. The UNRES effective energy
function is defined as a potential of mean force of
polypeptide chains in water, which has been subsequently
expanded into Kubo cluster cumulant functions, identified with
the respective energy terms. UNRES simulations have been used
with success in protein-structure prediction,
studying
protein-folding kinetics and free-energy landscapes as well as
to solve biological problems. Recently, we have implemented a
web server for coarse-grained simulations using the UNRES
force field. Local energy minimization, canonical molecular
dynamics simulations, replica exchange (REMD) and multiplexed
replica exchange molecular dynamics (MREMD) simulations can be
run with the current UNRES server. The user-supplied input
includes protein sequence and, optionally, restraints from
secondary-structure prediction or small x-ray scattering
(SAXS) data, and simulation type and parameters
which are selected or typed in. Oligomeric proteins,
as well as those containing D-amino-acid residues and
disulfide links, can be treated. The REMD/MREMD simulations
are followed by the weighted histogram analysis method (WHAM)
to compute the probabilities of conformations, thermodynamic
quantities and ensemble averages. Cluster analysis of the
ensemble at the desired temperature is carried out to
construct the final models which
are converted into all-atom representation. The UNRES server
is available at http://unres-server.chem.ug.edu.pl. This
website is free and open to all users and there is no login
requirement.
SPEAKER: Jayajit Das
(Nationwide Children’s Hospital and The Ohio State
University)
TITLE: Modeling
formation of biofilms by a bacterial pathogen in vitro and in
vivo
ABSTRACT:
Biofilms formed by nontypeable Haemophilus influenzae (NTHI)
bacteria play an important role in multiple respiratory tract
diseases. However, the mechanisms that underlie the emergence
of specific NTHI biofilm structures in vitro and in
vivo are unclear. We combined computational
analysis tools and in silico modeling rooted
in statistical physics with confocal imaging of NTHI biofilms
formed in vitro during static culture and in
the middle ears of chinchillas in order to identify mechanisms
that give rise to distinguishing morphological features.
SPEAKER:
Dawson, John (University of California-Davis)
TITLE: Are Two Blockers Better Than One: Investigating Whether
Beta Block Ameliorates hERG Blockade.
ABSTRACT: Beta-blocking drugs are a standard therapy for
the prevention of cardiac arrythmias. They bind to
beta-adrenergic receptors and attenuate downstream signaling
cascades, suppressing the sympathetic regulation of cardiac
output. Some prescribed beta-blockers also block hERG
potassium channel current (IKr), which can prolong the
repolarization phase of the cardiac action potential,
eliciting either beneficial anti-arrhythmic or potentially
deadly arrhythmogenic effects. Such is the case of dl-sotalol,
an antiarrhythmic medication representing a racemic mixture of
d- and l-sotalol, which have
similar hERG binding affinities. However, the l-stereoisomer
has substantially stronger non-selective beta-blocking
properties. In the infamous “Survival with Oral D-sotalol"
(SWORD) trial, researchers found that patients taking
d-sotalol exhibited a higher mortality rate than placebo due
to acquired arrhythmogenicity. It can be presumed that a
concomitant beta-blockade by l-sotalol in the racemic mixture
conveys a protective effect in dl-sotalol administration,
although this drug still possesses significant pro-arrhythmia
risks. The widely used beta-blocking drug propranolol also has
hERG binding affinity, but a better cardiac safety profile.
The precise molecular mechanisms of the combined effect of
hERG and beta-blocking activities of these and other drugs
remain undetermined. Elucidating them and devising
computational methods to predict their effects on cardiac
rhythm can provide a new paradigm for drug cardiac safety
testing and perhaps new avenues for rehabilitating drugs
otherwise deemed unsafe. Our laboratory employed
atomistic-scale molecular modeling and simulation techniques
to assess hERG- and beta-blockade for several drugs, including
dl-sotalol and propranolol. Here we present the development of
human beta-adrenergic receptor and hERG channel models and
their subsequent drug docking and molecular dynamics
simulations to identify key energetic and kinetic parameters
for their protein target interactions. We will implement these
parameters into a multi-scale functional model to predict a
combined effect of hERG and beta block on cardiac rhythm for
different drug doses and combinations.
SPEAKER: DeMarco, Kevin (University of
California-Davis)
TITLE: A Multiscale Model for in silico Prediction of
Drug-induced Arrhythmogenic Risk
ABSTRACT: We present a multi-scale computational
platform for the prediction of drug-induced arrhythmogenesis
resulting from blockade of the voltage gated potassium channel
encoded by the human Ether-à-go-go-related gene (hERG) in
ventricular cardiomyocytes. The hERG potassium channel
gives rise to a major repolarization current (IKr) in the
heart and is infamous for interacting with a diverse set of
drugs that can lead to acquired long QT syndrome (LQTS) in
humans, which is a surrogate indicator for potentially deadly
arrhythmias. However, not all hERG blockers are proarrhythmic,
and the underlying mechanisms relating hERG blockade and LQTS
to emergent arrhythmogenicity are not well understood. Our
multi-scale simulation framework utilizes predictions from
atomistic hERG-drug simulations to generate functional kinetic
model parameters that capture the dynamic electrophysiological
effects of IKr block. We focused on a prototypical hERG
blocking drug, dofetilide, which has a high proarrhythmic
risk. Umbrella sampling molecular dynamics simulations of
dofetilide in different ionization states with our open hERG
model were used to compute potentials of mean force and
diffusion coefficient profiles, from which we estimated drug
binding affinities (Kd), as well as entry and egress rates.
These data were then integrated into higher-order functional
kinetic models at the channel, cell and tissue scales to
identify the arrhythmia vulnerability mechanisms underlying
emergent arrhythmogenic phenotypes, arising from molecular
level hERG-drug interactions. We demonstrated this
proof-of-concept multi-scale model by predicting differential
arrhythmia markers, such as Triangulation, Reverse use
dependence, beat-to-beat Instability of action potential
duration, and temporal and spatial action potential duration
Dispersion (TRIaD), upon application of dofetilide.
SPEAKER: Emigh, Aiyana (University of
California-Davis)
TITLE: Predicting Arrhythmogenicity: Structural Modeling
of Safe and Unsafe hERG Blockers
ABSTRACT: Human Ether-a-go-go-Related Gene (hERG)
encodes a potassium-selective voltage-gated ion channel,
KV11.1, essential for normal electrical activity in the heart.
hERG mutations and blockage of the
channel pore by drugs can cause long QT syndrome (LQTS) that
predisposes individuals to arrhythmia and puts them at risk
for stroke or sudden cardiac arrest. A major problem in
antiarrhythmic drug therapies as well as drug development in
general is the proclivity for many drugs and drug candidates
to promote fatal arrhythmias through hERG blockade. However,
not all hERG blocking drugs are pro-arrhythmic, and their
differential affinities to discrete channel states and/or
their state stability modulations have been suggested to
contribute to arrhythmogenicity. In this study, we used
Rosetta electron density refinement, loop rebuilding and
homology modeling approaches to build complete structural
models of wild-type and mutant hERG channels in open and
closed states based on cryo-electron microscopy structures of
hERG (pdb id: 5VA2) and the homologous EAG1 channel(pdb id: 5K7L), respectively.
Here we present RosettaLigand molecular docking results for
drug interactions with hERG in open and closed states. We
observed that drugs with different pro-arrhythmia risks like
dofetilide and nifekalant have distinct binding poses and
affinities in closed and open hERG cavities, which might
affect channel gating. Also, Tyr652 residue in hERG
pore-lining S6 helix was found to be crucial for stabilizing
drug - channel binding in agreement with experiment. Our
results provide structural insights into molecular mechanisms
of state-dependent drug interactions with hERG that play a key
role in differentiating safe and unsafe hERG blockers.
SPEAKER: Espana, Guido (University of Notre
Dame)
TITLE: Using an agent-based model of dengue virus
transmission to estimate the impact of vaccination strategies
in different settings
ABSTRACT: Dengue is the fastest spreading vector-borne
disease worldwide causing around 100 mil- lion cases every
year. Dengue virus transmission is highly heterogeneous and its limited to temperate climates where
its main vector of transmission, the Aedes aegypti mosquitoes,
can survive. Even though a vaccine (CYD-TDV) has been develop against dengue, its efficacy is
heterogeneous in the population. Hence, the World Health
Organization has recommended to use
serological testing prior to vaccination to administer the
vaccine only to individuals that would benefit from it. To
estimate the public health impact of the dengue vaccine,
mathematical models should include sources of heterogeneity
that could affect indirect protection from the vaccine. We
used a stochastic, agent-based model to realistically account
for the various factors of heterogeneity involved in dengue
virus trans- mission. The model parameters were calibrated
using more than a decade of empirical studies in the city of
Iquitos, Peru. We used this model in various settings to
simulate 40 years of dengue virus transmission, followed by 10
more years of routine vaccination. We calculated the potential
benefits from the vaccine assuming different levels of the
accu- racy of serological screening to detect the target
population. In addition, we estimated the cost-effectiveness
of these measures for three specific countries: the
Philippines, Brazil, and Puerto Rico. Our results suggest that
this vaccination strategy could benefit some coun- tries,
particularly those with high transmission intensity. From a
public payer perspective, pre-vaccination screening was
cost-effective in scenarios with high screening sensitivity.
In conclusion, vaccination with CYD-TDV following serological
screening could have a posi- tive impact in certain
epidemiological settings, provided that screening is highly
specific (to reduce individual harm), at least moderately
sensitive (to increase public health benefits), and
inexpensive.
SPEAKER: Fidelis, Krzysztof
(University of California-David)
TITLE: “Sequence-structure
relationship
in proteins: An amino acid structural neighborhood
perspective”
ABSTRACT: Recent CASP13
assessments point to dramatic advancements in the protein
residue-residue contact prediction and in modeling of protein
structure itself. The progress is especially noteworthy for
proteins with no available structural templates, where until
now the most reliable technique, comparative modeling, cannot
be used. But how do the recent advances relate to the original
Anfinsenian stating of the protein
folding problem over 50 years ago, where the amino
acid sequence itself purportedly contains all the necessary
information to answer the folding question? In the current
methods, what is the role of the co-evolution based contact
prediction and how to explain that of the artificial
intelligence /deep learning? We take initial steps to
answering these questions by looking at levels of the amino
acid sequence similarity in local structures of proteins that
are not necessarily related by homology. Using our previously
developed formalism of structure descriptors, we examine
levels of local structure ubiquity among protein folds. Then,
for each local structure type, we assess whether sequence
signal can be seen between representatives taken from
different protein folds. We also examine whether local
structure libraries can be used to model any new protein, i.e.
if the structural coverage by library members is sufficient to
assemble a complete model. Finally, we present an analysis of
the CASP13 template-free and template-based modeling targets
in terms of library coverage and the levels of sequence signal
measured to the nearest correctly aligned library member.
SPEAKER:
Gnona, Komla (Nationwide Children’s Hospital)
TITLE: Developing genetic biomarkers for polygenic conditions:
An Application to Neonatal Complications in Preterm Infants
ABSTRACT: Preterm birth is the leading cause of
mortality in young children with more than a million deaths
per year worldwide arising from neonatal complications (NC).
Our most recent work (and the work of others) suggests that NC
is heritable and the genetic burden of as many as 5,000 genes
may be influencing NC. However, to date, no one has identified
a single gene for this condition. This suggests that most of
the genes involved in NC have small effects. As such, we
hypothesize that which genes get selected for biomarker construction don’t really matter. To
test this hypothesis, we (1) examined three different
prediction approaches: random forest, lasso, and polygenic
risk scores, applied to the whole-exome data of 131 preterm
infants; (2) looked at the different types of variants (e.g.
exonic, intronic, regulatory) and the average burden in two
disjoint gene sets: all genes with p<0.2 (ie, weakly
associated) and all genes with p>0.2 (ie, non-associated).
We found that most of the associated genes tend to be
intronic, suggesting that their biological effects are likely
small, and confirming that NC is polygenetic. We also found
that all the AUCs from the various prediction algorithms look
the same, even though vastly different computational
approaches were used. In conclusion, genetic biomarkers for
polygenic traits can be improved without knowing exactly the
genes involved, and replicating an association before building
a genetic biomarker for such traits may not be prudent.
SPEAKER: Jernigan,
Robert (Iowa State University)
TITLE: Using High Order Coevolution Correlations to Identify
Sites for Compensating Mutations to Rescue Function
ABSTRACT: Dense packing within globular proteins manifests
itself in strong interdependencies for amino acid
substitutions leading to strong correlations among them. The
available sequence data is growing rapidly, and it can provide
substantial important information regarding function and
design. We use pair correlations in the multiple sequence
alignments to improve protein sequence matching to bring
sequence matches into congruence with sequence matches. We use
mutual information to obtain the pairwise correlation and
apply a method called symmetrized differential interaction
information to measure the triplet correlation from the
available large protein sequence information. Combining the
results from these two methods, we are able to extract the top
correlated amino acids acting as hubs for allosteric
communication pathways. This method is applied to a variant of
Bruton’s tyrosine kinase (BTK) linked to a disease, severe
XLA. We previously showed that this deleterious mutant
adversely affects the functional dynamics of the BTK kinase
domain, and thus presumably the function. In another approach
we utilize comparative cross-species comparisons in sequence
alignments to identify the occurrence of a specific disease
mutant in the native sequence of another organism, and uncover
what other substitutions are likely to compensate for this
change. We suggest this may be a way to identify candidate
substitutions in the sites most strongly correlated with the
disease mutant to restore the native dynamics in the presence
of the deleterious mutation.
SPEAKER: Jiang, Xianli (University of
Texas-Dallas)
TITLE: Unraveling and designing signal-response connections
using direct couplings from a global coevolutionary model
ABSTRACT: Coevolution plays a fundamental role in
determining folding, structure, interactions and functionality
of proteins. Structurally or functionally related residues
coevolve during the evolutionary history due to selective
pressure. Direct coupling analysis (DCA) model for
coevolutionary analysis has demonstrated outstanding
performances in predicting coupled residues key to structural
or functional dynamics.We have
used coevolutionary traits among LacI homologs to develop a
model for predicting compatibility between an environmental
sensing module (ESM) and a DNA recognition module (DRM). The
inter-modular coevolutionary signature arises from the
constraints on allosteric dynamics, which involve specific
residue-residue associations. Engineering allosteric
transcriptional repressors containing distinct but compatible
ESM and DRM has the potential to flexibly construct new
connections for environmental signals and biological
responses. Our predictions accurately agree with experiments
for a collection of engineered repressors. The engineered
repressors are further explored in synthetic biology by
developing a system of genetic circuits called multiple toggle
switches with a master OFF signal. Coevolutionary model also
has unraveled signal-response connections in two- component
system where direct couplings have been used to evaluate
binding specificity of protein-protein interactions. The mode
accurately predicts partnerships not only between cognate
pairs but also cross-talk, leading
to a better understanding of specific phenotypes in
Enterococcus Faecalis in response to antibiotic treatments.
SPEAKER: Kool, Daniel (Iowa State
University)
TITLE: Investigating the Changes in Amino Acid
Properties in the Evolutionary and Multi-scale Context
ABSTRACT: Understanding protein mutations is critical
for comprehending evolution, drug development, personalized
medicine, genomics and much more throughout biology.
Classification here is performed by using features, which can
be generated/extracted from data, or chosen
by an investigator. One kind of feature set showing initial
promise not only for classification but also finding out what
the basis is for each classification utilizes the physical
properties of amino acids. It has been shown that there are
periodicities and patterns in protein sequences when they are
encoded as these physical property values. An encoding of a
sequence where each residue becomes a vector of the amino
acid’s physical property values provides much more information
for sequence similarity measures and can serve as feature
components for classifiers. Once a classifier is constructed
and trained, it can then be analyzed for feature importance.
One of the best methods for this is the permutation importance
algorithm. For each of the features, the values are randomly
permuted, and then the prediction error is recalculated. The
change in the error determines the feature’s importance for
any specific model, and this technique can be used on any
trained classification or regression model. Classification of
protein mutants based on physical properties combined with
permutation importance can reveal information about why or how
each mutation leads to changes in structure, dynamics or
function. Extensions include investigating the correlations
between properties at residues seen to be
correlated in the sequence alignments. This
method can be used on any scale, with any set of features,
making it a potentially powerful tool for protein analysis.
SPEAKER:
Grudinin, Sergei (Inria-CNRS)
TITLE:
Coarse-graining protein dynamics and representation for
integrative structural bioinformatics
ABSTRACT:
Our team develops
novel algorithms for integrative structural bioinformatics,
which are based on a combination of physics-based and
knowledge-based approaches. I will briefly present
computational pipelines developed in our team. I will also
describe in more detail 3 coarse-graining approaches.
SPEAKER: Kernik, Divya
(University of California-Davis)
TITLE: A computational framework to predict
mechanisms of phenotypic variability and disease severity in
iPSC-CMs
ABSTRACT: There is a profound need to develop a strategy to
predict patient-to-patient vulnerability in the emergence of
cardiac arrhythmia. A promising in vitro method
to address patient- specific proclivity to cardiac disease
utilizes induced pluripotent stem cell-derived cardiomyocytes
(iPSC-CMs). A major strength of this approach is that iPSC-CMs
contain donor genetic information and therefore capture
patient-specific genotype-phenotype relationships. A cited
detriment of iPSC-CMs is the cell-to-cell variability observed
in electrical activity. We postulated, however, that
cell-to-cell variability may
constitute a strength when appropriately utilized in a
computational framework to build in silico cell
populations that can be employed to identify phenotypic
mechanisms and pinpoint key sensitive parameters. Thus, we
have exploited variation in experimental data across multiple
laboratories to develop a computational framework to
investigate subcellular phenotypic mechanisms in healthy
iPSC-CMs. The resulting population of cellular models predicts
robust inter-subject variability in iPSC-CMs and recapitulates
the experimentally observed range of whole-cell behaviors. We
used this population-based framework to analyze a panel of
genetic mutations related to Long QT Syndrome 1 (LQT1).
Furthermore, we used our iPSC-CM model population to predict
the severity of LQT1 mutations and explore patient-specific
susceptibility to LQT1. LQT
mutations are known to have vastly different cardiac effects
in different patients, and these phenotypic differences are
recapitulated in our model population.
SPEAKER: Khade, Pranav (Iowa
State University)
TITLE: Using alpha shapes to characterize protein packing and
capture the multiscale aspects of allostery
ABSTRACT: There are only limited methods available to study
the global motions of the protein such as their hinge motions
and shear motions. These motions take place over a broad range
of time scales, from microseconds to seconds; however,
molecular dynamics methods can only model easily the motions
occurring on the time scale from picoseconds to microseconds,
and in addition such simulations require that replicas be run.
Thus, extracting the meaningful slow motions is difficult.
Hence, there is a need to explore other ways to utilize the
protein structures to model/ predict the global/large scale
motions of the protein. The important motions depend on the
protein packing as a multiscale phenomenon
which can influence the global or local motions in
the proteins. In order to model the protein packing, an
efficient, robust and simple mathematical method is needed. In
this study, we have explored alpha shapes (a subset of
Delaunay tessellations) for the protein backbone coordinates
as a model of protein packing. We demonstrate that the method
can predict the protein hinges which
are responsible for the global motions of the proteins. The
method is named PACKMAN (PACking and Motion ANalyses). From a
literature survey for randomly selected protein structures and
another 367 protein structure pairs having known open and
closed conformations, PACKMAN is able to predict hinge
locations accurately on the proteins for both the open and
closed conformations outperforming existing hinge predicting
methods that usually require either the open form or both
conformations to predict the protein hinges. The successful
implementation of the mathematical method to model a
multiscale phenomenon such as protein packing to predict the
hotspots of the global/large scale motions in proteins shows
promise for the further exploration of other types of protein
and supramolecular dynamics.
SPEAKER: Klaus, Colin (The Ohio State
University)
TITLE: Multiscale Approaches in Visual Transduction: A
Novel Homogenized Diffusion Model for Cone Photoreceptors in
Dim and Bright Light
ABSTRACT: Mammals have two types of photoreceptors, rods
and cones. While rods are exceptionally sensitive and mediate
vision at very low illumination levels, cones operate in
daylight and are responsible for the bulk of visual
perception. Through the mathematical techniques of
homogenization and concentrated capacity, a novel diffusion
model for cone visual transduction is presented along with
preliminary sensitivity analysis of model parameters. This
work also adapts a recently published dim light model of cone
transduction to the case of high intensity light. This study
is an interdisciplinary undertaking by mathematicians and
pharmacologists at The Ohio State University, CNR IT, and
Vanderbilt University.
SPEAKER: Kloczkowski,
Andrzej (Nationwide Children’s Hospital & The Ohio State
University
TITLE: Modeling structure, stability and dynamics
of proteins and protein aggregates
ABSTRACT: Recent progress in
modeling structure and dynamics of proteins and protein
aggregates will be reviewed. Recent advancements in structure
prediction such as development of better potentials and force-fields (including multibody
potentials), and improved modeling of free energies will be
presented. We significantly improved protein structure
evaluations by considering the effects of amino acid variants
on protein stability, and we have shown that the outliers in
stability are typically aberrant proteins. Recently, we
have made significant progress in understanding protein
stability and dynamics by computing protein free energies
extracted from structures to account for the high packing
densities in proteins, including important novel evaluations
of protein entropies. Preliminary results show large
improvements over previous potentials for assessing protein
stabilities. Our results demonstrate that there are
substantial gains in specificity from combining the sequence
with structural and protein dynamic data. These developments
may significantly impact the advancement of
precision/personalized medicine.
SPEAKER: Kouza, Maksim (Nationwide
Children’s Hospital)
TITLE: Assessing the Rate of Fibril Formation using
Steered Molecular Dynamics Simulations
ABSTRACT: Fibril formation resulting from protein
misfolding and aggregation is a hallmark of several
neurodegenerative diseases such as Alzheimer’s and Parkinson’s
diseases. Despite much progress in the understanding of the
protein aggregation process, the factors governing fibril
formation rates and fibril stability have not been fully
understood. Using all-atom explicit solvent molecular dynamics
simulations with the GROMOS43a1 force field for full-length
amyloid beta peptides Aβ40 and Aβ42 and truncated peptides, we
demonstrated that kinetic stability can be accessed via
mechanical stability in such a way that the higher the
mechanical stability or the kinetic stability, the faster the
fibril formation1,2. This result
opens up a new way for predicting fibril formation rates based
on mechanical stability that may be easily estimated by
steered molecular dynamics.
SPEAKER:
Joachimiak, Andrzej (University of Chicago)
TITLE: Addressing Ligand Binding
Promiscuity
ABSTRACT:
Enzymes
are typically described as molecular machines selectively
recognizing a ligand and cofactors and performing very
specific chemical transformations. But in reality, this is not
even close to being accurate. Recent studies showed that
proteins are capable of recognizing multiple substrates and
enzymes can perform different, sometimes unrelated chemical
reactions. These
activities can provide the starting point for evolution of
new functions and pathways. Promiscuous activities
available in nature appear to be very common and extensive.
This has important practical implications as we are facing the
rise of antibiotic resistance and enzymes rapidly evolving to
recognize and neutralize new “synthetic” derivatives of
antibiotic. Promiscuous
enzymes and evolution of secondary metabolisms is important
for synthetic biology efforts to construct novel pathways
using catalysts derived from promiscuous enzymes via
directed evolution. Structure-guided mutagenesis of active
sites at the substrate-binding pocket is responsible for
altering the specificity and promiscuity toward substrates
and the diversity of products.
High-resolution structures
obtained through x-ray crystallography can reveal specific
protein-ligand interactions and when combined with functional
assays and numerical simulation can construct a model of the
enzyme catalytic pathway and account for recognition
structurally unrelated ligands.
This can have pronounced impact on designing new,
better enzyme inhibitors and eventually drugs.
Several examples will be provided
including metallo-β-lactamases, inosine
monophosphate dehyrogenases, solute-binding proteins of ABC
transporters and transcription factors. Structures
of b-lactamases
belonging to four classes were obtained with the same
antibiotic and revealed differences in binding to their
respective active sites. The structures with hydrolyzed
antibiotics showed differences in the product binding to the
same enzyme. Inosine
monophosphate dehydrogenase binds multiple compounds and
was discovered to catalyze new chemical transformation. Solute-binding proteins of
ABC transporters can recognize multiple ligands and
transcription factors respond to different compounds.
This
project has been funded in whole or in part with Federal funds
from the National Institute of Allergy and Infectious
Diseases, National Institutes of Health, Department of Health
and Human Services, under Contracts No. HHSN272201700060C.
SPEAKER: Lichtarge, Olivier
(Baylor College of Medicine)
TITLE:The Skinny From All of Us to Each of
Us: A Calculus of
Fitness Suggests Simple Rules for Complex Diseases
ABSTRACT: The genotype to
phenotype relationship shapes evolution in the long run and
human health in all of us. Rooted in random genetic errors and
cryptic past selection forces, it seems intractably complex. Nevertheless, a
shift in perspective to the proper scale and resolution
suffices for basic laws of calculus and statistical mechanics
to describe genetic moves in fitness landscapes with simple
yet general rules. From cancer to Alzheimer’s disease, this
shift in perspective not only yields a synthesis between
molecular and population genetics but also reveals new disease
genes and plausible strategies for risk assessment and
clinical decisions tailored to each of us.
SPEAKER:
Ma, Qin (The Ohio State University)
TITLE: Analyzing Co-regulated Gene Modules
ABSTRACT: A gene’s expression is regulated by a set of
TRSs, including transcription factors (TFs), microRNAs, long
non-coding RNA, and epigenomic regulators. Genes co-regulated
by the same TRSs are collectively referred to as a
co-regulated gene module (CRGM). Accurately identifying CRGMs
is a key to understanding gene regulation and the underlying
TRSs and the prerequisite to discover the genetic and
epigenetic machinery that leading to
different cell types. Single-cell multi-omics sequencing
technology can detect both epigenetic and genetic information
from an individual cell. Intuitively, the joint analysis of
these data provides an unprecedented opportunity to predict
all the encoded CRGM at the single-cell level, which in turn,
can substantially improve the state-of-the-art performance in
elucidating cellular heterogeneity. The successful
identification of underlying cell-type-specific CRGM can
substantially improve the elucidation of heterogeneous gene
regulatory mechanisms, and determine the identity of cell
types in immunotherapy. It will also benefit the design and
assessment of therapeutics in complex diseases.
SPEAKER:
Miao,
Yinglong (University of Kansas)
TITLE: Gaussian
accelerated
molecular dynamics (GaMD): Bridging gaps in multiscale
simulations
ABSTRACT: Gaussian accelerated
molecular dynamics (GaMD) is a robust computational technique
that provides simultaneous unconstrained enhanced sampling and
free energy calculations of large biomolecules. GaMD works by
adding a harmonic boost potential to smooth biomolecular
potential energy surface and reduce the energy barriers.
Without the need to set predefined collective variables, GaMD
is advantageous for studying complex biological processes with
no constraints applied in the simulations. Furthermore,
because the boost potential follows a Gaussian distribution,
the original free energy profiles of biomolecules can be
recovered through cumulant expansion to the second order
(“Gaussian approximation”). GaMD allows us to more accurately
characterize biomolecular dynamics. Compared with conventional
molecular dynamics, GaMD accelerates biomolecular simulations
by orders of magnitude. This allows us to bridge gaps in
multiscale simulations of biomolecular dynamics. Here, new
method developments of GaMD and our recent applications of
GaMD in accelerated simulations of ligand binding,
peptide-protein binding and protein-protein/membrane
interactions will be presented.
SPEAKER: Pietrzak,
Maciej (The Ohio State University)
TITLE: Application of information-theoretical approaches
for the analysis of human immunome.
ABSTRACT: The cellular immunome (the cellular components
of the immune system) is a vastly complex and highly regulated
structure that protects against infection and preserves good
health. An ever-increasing number of cell types comprise the
immunome, and these are being defined through increasingly
complex patterns of antigen expression. Recent technological
advances over the last decade now permit high-dimensional
examination of the immunome cellular components, making
possible in-depth analysis of the cellular immune system. The
analytical frameworks for comprehensive assessment of immunome
profiles are not well established yet. The complexity of data
produced by the flow cytometric analysis of the human immunome
requires computational approach that would allow to detect not
only the large-scale changes in the dominant components of the
immunome, but, what seems more important, consistent
differences in the non-abundant components and relations
between them. In this talk, I will present the preliminary
results of a recent work on development of information-based
feature selection algorithm to detect alternations of human
immunome profiles.
SPEAKER: Rempala, Grzegorz (The
Ohio State University)
TITLE: Multiscale Stochastic
Models of Transcription/Translation
ABSTRACT: The formalism of stochastic reaction networks
(SRNs) provides building blocks for many models in modern
mathematical biology both at molecular and population levels
(e.g., gene transcription or epidemic outbreak). In
particular, the SRNs allow to naturally incorporate both delay and multi-scale
phenomena into computational models. In this talk I will
provide a brief overview of the applications of SRNs to
modeling basic molecular mechanisms emphasizing
some recent collaborative work on multi-scaling analysis
for simple stochastic gene transcription.
SPEAKER: Rohs, Remo
(University of Southern California)
TITLE: Multiscale
Modeling of Protein-DNA Binding Specificity
ABSTRACT: Many
structures of protein-DNA complexes have been solved and
high-throughput binding assays were developed because
structural biology and genomics researchers were equally
puzzled by the question of how proteins bind DNA with high
specificity. However, there was little communication between
these two fields of research. High-throughput DNA shape
prediction established a cross talk between both fields. The
primary goal of DNA shape analysis remains the quest for
mechanistic insights into protein-DNA readout modes based on
sequencing data without the need of structure determination. A
plethora of high-throughput sequencing data is available from
a variety of experimental approaches. In contrast, structural
biology, albeit being an atomic-resolution approach, often
reports the binding of a single protein to only a single DNA
target. A number of studies incorporated DNA shape features in
the quantitative modeling of binding specificities. These
studies emphasized the importance of interactions between
nucleotide positions within a binding site and its flanks,
although the definitions of DNA sequence versus shape still
differ in the structural biology and genomics fields. Next
steps discussed in the talk focus on understanding biological
phenomena such as purifying selection, additional layers of
binding specificity determinants such as DNA methylation and
histone modifications, and new computational approaches
including artificial intelligence and quantum computing.
SPEAKER: Saiz, Leonor
(University of California-Davis)
TITLE:
Multilevel
modeling
of cellular networks in biomedicine
ABSTRACT: One
of the main challenges of current biology is to integrate the
available genetic, biochemical, molecular, and structural
information into a physiologically relevant description of
cellular and supracellular processes. Computational modeling
has emerged as a promising tool for transforming molecular
detail into a more integrated form of understanding complex
behavior. In this talk, I will draw examples from the recent
work of my laboratory to discuss the
state-of-the-art on modeling of
biological processes at different temporal and spatial
scales, going from molecular interactions to the assembly of
macromolecular complexes on DNA and the stochastic dynamics
of the resulting gene regulation networks. I will discuss
our recent
results on the study of genetic
and signal transduction networks including gene regulation
by the RXR nuclear receptor and signalprocessing in the
TGF-beta pathway.
SPEAKER: Mateusz
Sikora (Max
Planck Institute of Biophysics)
TITLE: Truss-like
arrangement of cadherins is responsible for desmosome strength
ABSTRACT: Desmosomes are
long-lasting cell-cell junctions that endow mature tissues
with mechanical stability. The core of
the desmosomal adhesion is formed by cadherins, specialised
calciumdependent transmembrane proteins. Together
with other adaptor proteins, the cadherins connect cortices of
neighbouring cells. Robustness to external stress comes from a
particularly dense arrangement of cadherins, which form
characteristic electron-dense structures visible in EM
micrographs. However, the structural details of desmosomal
cadherin assemblies remain controversial despite their
relevance for various diseases. To address and resolve these
controversies, we performed large-scale molecular dynamics
simulations of a different 3D cadherin
arrangements in the desmosome. We found that only an
antiparallel, truss-like arrangement of cadherins can explain
both the mechanical robustness and the spacing observed
between plasma membranes in the desmosome. We validated our
predictions by cryo-electron tomography of the desmosomes from
mouse liver.
SPEAKER: Stewart,
William (Nationwide Children’s Hospital and The Ohio State
University)
TITLE:
Multi-scale
Modeling of Serial Measurements in Survival Data
ABSTRACT: An extremely
important quantity that is difficult to predict accurately is
1-year post-transplant survival of candidates on the national
waiting list for organ transplantation. Donor organs are a
precious human resource, and the demand remains higher than
the supply, especially for lungs. The current system for
allocating donor lungs is based on a measure (denoted LAS)
that balances post-transplant survival against medical need.
In general, the highest priority is given to the candidate
with the highest LAS. However, because donor organs become
available at random times in random geographical locations,
each candidate’s LAS is calculated
frequently. The calculations do not occur at scheduled times,
and the intervals between calculations are not regular. As
such, these serial measurements (ie, LAS trajectories) are
inherently noisy, but they may also contain new and relevant
information that is not necessarily captured by LAS at a
single time point. To test this hypothesis, we propose a
mathematical concept that seeks to improve predictions of
post-transplant survival using each candidate's LAS
trajectory. Ideally, one would like to integrate both the
cross-sectional and longitudinal data simultaneously.
SPEAKER: Vakser, Ilya
(University of Kansas)
TITLE:
Docking of protein models
ABSTRACT: Modeling is important for
structural characterization of macromolecular interactions.
The accuracy of modeling, in general, is still less than that
of the experimental approaches. To test the ability of docking
procedures to predict the structure of protein model-model
complexes, we built docking benchmark sets of protein models,
with accuracy assessed by the modeling rank, for protein
complexes from the Dockground
docking benchmark and GWIDD database. The sets are different
from previously generated ones, in which arrays of models were
generated according to RMSD from the native structure. The
current sets rather reflect the real case modeling/docking
scenario where the accuracy of the models is assessed by the
modeling procedure, with no reference to the native structure,
which would be unknown in practical applications. The sets are
available at http://dockground.compbio.ku.edu. The docking
success rate was correlated with the modeling parameters for
individual proteins. The exhaustive benchmarking was also done
for protein pairs in which proteins are
modeled at different levels of accuracy (e.g. low accuracy
"receptor" and high accuracy "ligand"), again
reflecting the real-case scenario of practical applications.
The results show that docking techniques are applicable to
protein models in a broad range of modeling accuracy. The
study provides guidelines for practical applications of
docking to protein models.
SPEAKER: Vargas
Bernal, Esteban (The Ohio State University)
TITLE: Relating Eulerian and Lagrangian
spatial models for vector-host disease dynamics through a
fundamental matrix
ABSTRACT: We explore the
relationship between Eulerian and Lagrangian approaches for
modeling movement in vector-borne diseases for discrete
space. In the Eulerian approach we account for the movement
of hosts explicitly through movement rates captured by a
Laplacian matrix L. In the Lagrangian approach we only
account for the proportion of time that individuals spend in
foreign patches through a residence matrix P. We establish a
relationship between an Eulerian
model and a Lagrangian model for the hosts in terms of the
matrices L and P. We explore the consistency of this
relationship and in each scenario (inconsistent and
consistent) and we look at the comparison of quantities like
final outbreak size and basic reproduction number for
reasonable parameters. In the case of two-patches model, we
give an insight on how to relate the residence times and the
flow rates according to how big the removal rates are with
respect to the movement rates. We also observe how even in
the inconsistent case, we could get similar values of the
basic reproduction number and final outbreak size.
