2 February 2005 2 February 2015 COSBI is a bioinformatics research center operating in the fields of systems nutrition and systems pharmacology
2 February 2005
2 February 2015
COSBI is a bioinformatics research center operating in the fields of systems nutrition and systems pharmacology
COSBI headquarters located in a historical tobacco factory
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 3
Paolo ColliniUniversity of Trento
Corrado PriamiUniversity of TrentoPresident & CEO
Pierpaolo DeganoUniversity of Pisa
Assembly of Parties
Structure
Opening
Advisory Board
Board of Directors
Pier Paolo Di FioreIFOM
Leroy HoodInstitute forSystems Biology
Michael MüllerUniversity of East Anglia
Previous AB members:
Marvin Cassman, David Harel, Manuel Peitsch, Judith Armitage, Gianfranco Balbo, John Heath, John Tyson
AB role:scientific advice and evaluation
2 February 2005the signature
7 December 2005opening with a message from the President of Italian Republic Ciampi
2 April 2006scientific opening
Jim KarkaniasMicrosoft Corporation
Luca CardelliMicrosoft Research
Daron GreenMicrosoft Research
4 • COSBI - April 2015
COSBI works in the bioinformatics market by applying proprietary and public methods to systems pharmacology and systems nutrition to promote health with personalized medicine and nutrition in a collaborative effort with the international scientific community.
COSBI operates in the context of translational medicine and nutrition by integrating molecular and clinical, qualitative and quantitative, large scale and mechanistic data from different sources. Experimental data are continuously enriched and validated with scientific literature search.
Expertise
Metabolic disorders:obesity, diabetes, metabolic syndromeNeurodegenerative disorders:Alzheimer’s, dementia, autismData types:genomics, proteomics, transcriptomics, lipidomics,metabolomics, clinical markers, diet, lifestyle, physiologicalAnalyses:data aggregation and exploration, clustering, topological and functional network analysis, literature andtext mining, machine learning, stochastic and deterministic simulation, data visualization
Some indexes
2010 2011 2012 2013Avg IF 3.59 3.62 4.54 4.3Journals 19 21 26 24TOT PUB 40 46 55 36Invited talks 19 23 29 25Man Months 294 225 214 185
WHO
measurements
data productionand collection
WHAT
module identification
network analysis
mechanistic details
modeling
WHY WHERE
simulation
scenario generation
WHEN
pred
iction
and
cont
rol
analy
sis in
terp
retat
ionhy
poth
esis
gene
ration
prediction and control
functional annotation + low throughput lab work
Molecular understanding of (nutritional) diseases and health
MECHANISTIC DYNAMIC
CORRELATION STATIC
clustering stratification
multisource, omicsdata aggregation
and analysis
HOW
H2OH+
Glycine GlycineCO2NH3
Serine
NAD(P)H
NAD(P)+
1037 963342 902969353959 9664 31095963 87799 26101585310 3926 86678966 89567 2810158696 849508767 10973 91645 2598588562 589458824 10950 92684 269838858 8 5940 90078968 916768 2110138896 6 495390159 8956 89067 2310089126 759479038 5936 86046 2400791586 4990 9407 5932 85755 249948948 501019928 5915 87685 261006900 56010089341937 89557 301008902 6 4397394175 938 9147 829991909 7 40984940 4
1037 963966 973 950 968 956 936 932915937938
clinical and physiological data
omics and molecular data
qua
litat
ive,
larg
e sc
ale
dat
a
quantitative, sm
all scale data
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 5
COSBI applies network analysis to identify diagnostic and prognostic biomarkers. Phenotypes and omics data analysis select the biological network that drives the biological process. These networks are simulated to elucidate the molecular mechanisms and to choose potential drug/micronutrient targets.
Simulation of molecular models is also used to determine the parameters of clinical and physiological models.
Important visits
Strategic partnerships
18 June 2008Turing Award winnerRobin Milner
6 November 2008Nobel Prize winner Sydney Brenner
23 June 2009Turing Award winnerTony Hoare
DATA
ClinicalMarkers
GenomeProteome
Metabolome
Scientificliterature
MetabolicNetwork Protein
ProteinInteraction
Network
GeneRegulatory
Network
Target IdentificationDrug RepositioningModule Identification And Raking
Network Analysis (Topology+data Driven)Data MiningEnrichment Analysis
In-silico experimentsWhat-If analysisPrediction and controlPk/PdDose-regimen analysis
Stochastic, Deterministic, Hybrid SimulationODE, Language-based ModelingInference Procedures/fitting
OntologiesOmim
Drug-banks ODEChemical reactions
Languages Inference
Modules
FittingKinetics
Genomics
OU
TP
UT
INP
UT
ME
TH
OD
S
Multiomics(Integrated Data Analysis)
Signatures
DiagnosisPrognosisHealth Measures
Patient Stratification
Quan
titat
iveHe
tero
gene
ous
Quali
tativ
eEn
viron
men
t
Metabolomics
Multi
-om
icsMu
lti-s
ource
Multi
-sca
leInd
ividu
al
DATA INTO CONTEXT DYNAMIC MODELS
Proteomics
Lipidomics
Diet
Microbiome
LifeStyle
Functional
Analysis
ORGAN-LEVELPHENOTYPE REPRESENTATIONS
MOLECULAR PROCESSES
Simulation of molecular interactionsHigh-level variables Equations
D: administered glucoseJ: jejunumR: delay cmptL: ileumG: plasma glucoseI: plasma insulin
S
J R L
G
IVI
VG
D
kjsS
krjS klrR
kgjJ kglL
GPROD
kxgilG
-kxlI
kigmax
?
TISSUEFORMATION
CELL-CELLINTERACTION
HORMONESIGNALING
CerSph
Lysosome
ER
aSMase
SM
SM
CERSph aCDase
Cell Membrane
CerS
Cytoplasm
Plasma
Nucleus
CDase
Salvage pathway
Endocyticvesicle
SK
SPPaseS1Plyase
Ethanolamine phosphate+
hexadecenal
3kdhSphpalmitoyl-CoA
L-Serine
dhSph
Acyl-CoAdhCer
Des1
SPT
DSR
CerSde novo synthesis
CDase
S1P
Mitochondria
CerS
Sph Cer SM
SMS
SMasenCDase
PC DAG
palmitoyl-CoA
CoA
Acot2
palmitate
SK
S1P
SphCer
Cer
SM
SMSphS1P
PC DAGSPPase
CDase
CDase
nSMase
SMS2
nSMase
SMS2
CFTRABC
S1PlyaseEthanolamine phosphate
+hexadecenal
PC DAG
Sphingomyelinase pathway
SMLacCer
GM3
GluCer
GluCer
Cer
GolgiApparatus
LacCerS
GM3S
FAPP2CPase
Cer1P Cer SM
SMS1
SMaseCerkSapC
PC DAG
GDaseGtase
CPE GalCerGLASMSr
Connecting de novoand sphingomyelinase
CERT
Gi/o
Gq
G12-13
PKCCa2+
PLC
PI3k
RAS
ERK
PKB/Akt
RAC
Migration,vascular tone, endothelial barrier
function, neural cell communication
Survival
Proliferation
S1PR2S1PR3
S1PR4
S1PR5
AC
cAMP
S1P signaling
RhO
Cdc42
S1PR1
JNK
TNF
TNFR
NF-kB
Inflammation
InsulinAction
IRS1
PI3k
Akt/PKB
PP2A
PKC
IRInsulin
Cer
Cer
IKK
CAPK
RAF1
MEK1
ERK1/ERK2
Cer signaling
Cer PP1
SR proteinsCaspase-9
BCL-X
Cer signaling
GM3
GM3
GM3 signaling
CatepsinD
Cav1
6 • COSBI - April 2015
Main journals
Nat communications, PloS ONE, Biol Letters, BMC Syst Biol, Bioinf, Drug Discovery Today, Nat Cell Biol Syst Biomed, Net Biology, J. Nutr Biochem, J Chem Phys, ACM Comp Surveys, Mol Nutr, Gene & Nutr, Clinical and Translational Gastroent, Food Res, J of Env Mgt, WIREs Syst Biol and Med
COSBI invented an innovative method to determine diagnostic and prognostic biomarkers validated in an international competition and industrial projects.
Gene Expression
Gene Ranking
A1
1
2 23
3
4 45
5
6 67
7
8
8
9 910
10
11 1112
12
13 1314
14
15
15
1616
17
17
1
1
2 23
3
4
4
5
5
6 67
7
8 89
9
10
11
11
12 1213
13
1415
15
16
17
17
1
1
2 23
3
4 45
5
667
7
8 89
9
10 1011
11
12
12
13 1314
14
15
15
16
16
17 17
B C A B C
Signature Extraction
A B C
SignatureComparison
d(A,C)2
410
1521013
162
4
11
16
15
816
17
914
3
59
17
A
B
C
2
410
15
21013
16
2
4
11
16
15
816
17
914
3
59
17
MapConstruction
A
B C
10
14
16
d(A,B) d(B,
C)AUTISM
MULTIPLE SCLEROSIS
Visual cortex
Posterior Cingulate Cortex
Superior Frontal Gyrus
Hippocampus
Medial temporal Gyrus
Enthorinal cortex
Microarrays from laser microdissected neurons
CONTROL
ACETAMINOPHEN (100mg/Kg)
6h
24h
3d
7d
ACETAMINOPHEN (1250mg/Kg)
High Dose 6 Hours
High Dose 24 Hours
High Dose 3 Days
Low Dose 6 Hours
Control
mRNA from liver in rats
GSE32891 (FDA, Je erson, AR) mRNA from L signature size: 50+50, top 20% edges
GSE37772 mRNA from lymphoblast cell lines derived from 386 individuals of 196 Simons Simplex Families Signature size: 25+25, top 20% edges
GSE37772 mRNA from lymphoblast cell lines derived from 386 individuals of 196 Simons Simplex Families Signature size: 25+25, top 20% edges
Main prizes
27 February 2009COSBI wins the competition Formal Methods for MolecularBiology with 22 participants
30 November 2010The President of the Italian Republic Napolitano appointsCOSBI with a medal for its results in the first 5 years of activity
2 October 2012COSBI wins second place over 52 participants in theinternational competition SBVImprover in Boston to determine biomarkers. COSBI is first in the sub-competition on multiple sclerosis
11-13 June 2014COSBI wins the first prize atSIBBM 2014 for its studies onneurodegenerative dementia
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 7
Main events and seminars
COSBI integrates multiple data types to correlateneurodegenerative diseases.
Similar methods have been used to identify Alzheimer’s biomarkers.
Information theory is used to determine the activity levelof biological networks.
Alzheimer relevant genes
Network analysis
mRNA expression
SNPs
Module genes
Modules in HPRD PPI network
GO terms(Biological Processes)
Functional annotation analysis
Identification of AD candidate genes and/or Biomarkers
Significantly enriched modules
Drug targets
OMIM genes
Inputs Processing Output
Identify modulesconnecting gene sets and receptors/transporters
Gene set
Interaction network
List of receptors/transporters
DEGs Fold change of DEGs
Pruning sub-network between selected receptors/transportersand gene set products
Topology
Node activity score and cellular functions+ p-value
Networkactivity score
Information theory-basedcomputation of node activity levels
Disease genes
Network reconstructionProtein-ProiteinInteraction network Shared node identification
GO terms Biological ProcessesAnalysus of specific GO term-associated genes
Network analysis FunctionalannotationanalysisSignificantly
enriched modules
OMIM genesHuntington, Prion, Frontotemporal dementia, Alzheimer’s, ALS, Friedreich ataxia, Lewy BD, Parkinson, SMA,Glioblastoma
follows>
Converging Sciences 2006
Biology without Borders 2007
BioComplex 2008
Merging Knowledge 2010
7 December 2005Leroy HoodSeminar
17 February 2011Larry WallSeminar
8 • COSBI - April 2015
COSBI is setting the state of the art in modeling languages and simulation algorithms.
Languages
COSBI in Scientific International Boards
Scientific boards: Fondazione Veronesi, EU ISTAG - Future and Emerging Tecnologies Unit, CRUI - programmi europei, MT-LAB.Review panels: MIUR, EU-IST FET, Ireland Science Foundation, Genome Canada, BBSRC, The Royal Society UK, European Research Council, Medical Research Council UK, Council of Earth and Life Sciences NL, BMBF D, The Netherlands Organisation for Research, Czech Science Foundation, Research Council of Lithuania, Austrian Science Fund.
RSSA: a new faster, exact simulation algorithm
Case
Stud
ies
Implementation
Stochasticpi-calculus1995
Beta binders2004
BlenX2008
L2012
COSBI is devoting a great deal of effort in making simulation accessible through user-friendly graphical interface and minimal languages.
SHMT
H2O
H+
NADP+
NADP+
NADPH
NADPH
NADP+
NADPH
DHFR
TYMS
THF
10f-THF
Formate
5,10 CH=THF
5,10 CH2-THFDHF
5m-THF
Met
SAM
SAH
Hcy
Methylation
Sumoylation
MTHFD1(FTHFS)
MTHFR
MTR
ADP+Pi
ATP
dUMP
Glycine
Serine
Purinesynthesis
CYTOSOLNUCLEUS
MTHFD1(MTHFC)
MTHFD1(MTHFD)
dTMP
sumoSHMT
NADP+
NADPH sumoDHFR
5,10 CH2-THF
THF
DHF
dUMP
Glycine
Serine
dTMP
Thymidylate Biosynthesis
sumoTYMS
[steps = 5000, delta = 0.2]let CYCBT: bproc = #(x,CYCBT)[ nil ];when(CYCBT:: d_dtCYCBT_1) new(1); when(CYCBT:: d_dtCYCBT_2) delete(1);when(CYCBT:: d_dtCYCBT_3) delete(1); when(CYCBT:: d_dtCYCBT_4) delete(1);let CDH1: bproc = #(y,CDH1)[ nil ]; let CDH1_IN : bproc = #(y_in,CDH1_IN) [ nil ];when(CDH1_IN :: d_dtCDH1_1 ) split(Nil, CDH1); when(CDH1_IN :: d_dtCDH1_2 ) split(Nil, CDH1);when(CDH1 :: d_dtCDH1_3 ) split(Nil, CDH1_IN); when(CDH1 :: d_dtCDH1_4 ) split(Nil, CDH1_IN);let CDC20_IN : bproc = #(a,CDC20_IN)[ nil ]; let CDC20_A : bproc = #(a,CDC20_A)[ nil ];when(CDC20_IN :: d_dtCDC20_IN_1 ) new(1); when(CDC20_IN :: d_dtCDC20_IN_2 ) new(1);when(CDC20_IN :: d_dtCDC20_IN_5 ) delete(1); when(CDC20_IN :: d_dtCDC20_IN_4 ) split(Nil, CDC20_A);when(CDC20_A :: d_dtCDC20_A_2) split(Nil,CDC20_IN); when(CDC20_A :: d_dtCDC20_A_3) delete(1);
A+B C
C+D
E G
G+H-E
L
-F
E
O43680
P12830Q99750P15173
O60682P52945P63279
Q16559Q6GYQ0Q7RTS1
Q02535
Q86U70
Q8TE12P11912
Q02363P41970
P50553Q8N436
P55036Q9NQ33
P49137
O00233
O95273
P0C1Z6
P41134Q99081
Q92858
Q02575Q16644
P01009
P13349O60663
Q15672P49639Q92878
P23409
Q9UL17P49711
P05113
P15260O60216
P17947
Q13546
P12956P49715
Q96S59
P17542P12980
P62158Q13309
P17676
Q13526
P10275
Q05516
Q99933
Q9Y3F4P51587Q8TEW8Q9NQ69
Q7LFL8
Q96D21O75593
Q9UEE9Q6DJT9P49796
Q7Z3E2
Q92963
Q9UJU2Q13484Q9H0M0
Q9NZH6Q05066Q9Y692
Q9UJ37
P54792Q9UKN5P57729
P11802P78451 Q13042
Q9Y5J3
P56915
P35573
P46108
P29590Q15797
P01100P18075
P51522
Q9UM63P06737P49716
Q8N2W9
Q9UK99
Q9UIS9Q9NPG8P35658
Q9NQ87
Q9P2J3Q9UPQ4
P08621
Q16254
Q03112
Q9GZN2Q08174
Q14213
Q16665P36897
Q9NPF7
P36896
P15692Q86YA9
P06753
P15822Q12778Q96SV5
Q9ULV0P21980Q96DE7
Q9UBH0
Q9HDC5
P61244P98177
P78368
P05388
O14543Q6ZNA4
Q14974
O00308Q7Z3B3
P46734
P35908
Q9UBW7Q9UMN6
P09466Q8WZ47
Q9H2X6
Q13608
Q96RT1
P11926
Q13445
Q8NDD8O43294
Q547Q3O43318
O60318
Q9Y2K7
P17535
Q9UHL9
P62070
O60315Q8TAI7
Q13485P28749
Q9HCE7
P01375
P84022
P24158
Q9UI36
Q15459Q13501
Q86UC2
Q9H7M9
Q8N6I1
O43524
Q9NRA8
Q99814Q14155Q9H0E3
Q15052
O00231O15105
P30260
Q9HAU4
Q8WXH2P30291Q9Y483
Q9UM13Q4LDE5
P13645Q15583
Q86UD4
O95405
P41182
Q92793
Q09472
Q15149
Q15326
Q6ZRS2
Q96EP0P43243
Q96EY1
O75376
P10914
P05412
Q9Y265 Q9Y230
Q01196
Q06330
P40763
P15336
P17275P46531
P12757
Q13227
Q9BZK7
Q9UBP5
Q16342
P18615
Q6MZP7Q15393Q14995
Q9UHI6
P12755
P49759
Q15303
Q86T24
P02794
Q14511Q15406O75081P10244
O14646Q92828
P50402Q9UPN9 Q7L2E3 Q96EB6
O75446
P20393P13236O43439
P34932Q96T58Q9UKV0Q86Y13
Q13490
Q9Y6X2
P15923P06454P19876P45452Q86UQ4
Q9UBX0O75164
P07585Q13443P21810P20333
Q12933Q8N165Q15628P78536
P19438
P09603
Q9ULZ9
Q8WUI4Q93009
Q16576Q9UQL6
Q9NNW7
O60907P52732
Q13263
Syndrome of “a new symbol”+
Better = New functionalities
COSBI Style
4 November 2014Pier Giuseppe PelicciSeminar
ECEM 2011
SAC 2012
< Main events and seminars
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 9
Institutionalvisits
13 November 2006Minister Fabio Mussi
2 April 2008United States Ambassador in Italy Ronald Spogli
19 May 2009President of PAT Lorenzo Dellai
COSBI develops software prototypes to apply the analytical methods that were designed and implemented for ad hoc biological problems.
COSBI has also developed COSBI LAB, a professional environment to model and simulate molecular biological processes.
Software prototypes
SCUDO is a tool for clustering gene expression profiles for diagnostic purposes using a new type of rank-based signatures
SCUDOL
An imperative, domain specific language to stochastically simulate biological systems
NASFinder
The Network Activity Score Finder is a web service for topological and functional analysis of sub-networks connecting an omics-determined module
SICOMPRE
Simulation-based, qualitatitve and quantitative prediction of protein complexes
BioNetMotion
BioNetMotion provides dynamic and network-based visualization of time course omic data
GENER
Gener is a tool for performing reductions on DNA-strands based on a strand-displacement algebra
LIME
Language Interface for individual-based modeling of ecosystem dynamics
WALDO
Waldo Reaction-based tool for easily modeling and simulating biological systems
BETAWB
BlenX-based tools to represent and simulate biological entities and their interactions
KINFER
Estimates both structural and nuisance model parameters from time-series data of reagents abundance
REDI
Simulates non-homogeneous and anisotropic stochastic di�usion of molecules
RSSA
Rejection-based Stochastic Simulation
COSBI LAB MODEL
estSestS
COMPONENT SITE
EST estSi
ERa
COSBI LAB SIMULATION
COSBI LAB PLOT
COSBI LAB GRAPH
COSBI LAB PLOT MATRIX
MOdEL SIMuLatIOn PLOt GraPh PLOt MatrIx
10 • COSBI - April 2015
Corporate visits
Main customers
People
technology transfer
7 December 2005 and5 June 2009Rick Rashid, SVP Microsoft Research
Nestlé Institute of Health SciencesPurinaUniversità di VeronaGlaxoSmithKlineSanofiUniversità di ParmaIEOAutifonySomaLogic
Avg age: 32Countries: 10Disciplines:Biology,ComputerScience,Bioinformatics,Mathematics,Ecology,Statistics,Engineering,Bioengineering
13 September 2007and 15 March 2013Tony Hey, VP Microsoft Research Connections
5 October 2012Ed Baetge, CEO Nestlé Institute of Health Sciences
COSBI refocused its activities from 2011 to be selfsustained by offering added-value scientific services in the fields of data analysis, modeling and simulation of biological process both to food and pharma industries and academic institutes and groups. This reorganization allowed COSBI to cover about 70% of its costs with commercial services and to change the structure of itsincome considerably.
0%
17,5%
35%
52,5%
70%
2010 2011 2012 2013 2014 2015
Public funds Academic projectsCommercial income Other income
the network of people grown at COSBI
Virginia Tech, University of Exeter, INRIA Paris, Universitad EAN, University of Lille, INRIA Rennes, RIKEN institute, University of Aalborg, Ecole Centrale Paris, Roche Diagnostics, Centro National de Biotecnologia Madrid, Navionics, University of Bolzano, Università di Trento, Fondazione Edmund Mach, University College London, Bax Energy, IMT Lucca, Accenture, ETH Zurich, University of Cincinnati, Università di Milano Bicocca, Trento RISE, Skype, SMC.
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 11
Contract research and servicesSystems pharmacology - Systems nutrition
1 biomarkers
these methods identify biological signatures that characterize a phenotype to classify samples. Biomarkers are diagnostic or prognostic, determine optimal groups in clinical trials, and drive the next steps in the pipeline
2 network selection
phenotypes or diseases select the gene, protein, metabolic, drug, microbiome, mixed network for the analysis
3 network analysis
omics and clinical data are used with topological indexes to identify modules of the network most significantly associated with the phenotype
4 functional analysis
omics, clinical data, knowledge from literature and DBs is used to identify the processes of themodules from step 3
5 simulation
modules are mapped into executable representations of the dynamics of the system to run virtual experiments
Each step of the pipeline produces significative results and it can be performed in isolation. The pipeline can start from each step and can continue until the desired results are obtained.
Each step of the pipeline applies a combination of proprietary methods and public methods to maximize the results. COSBI methods are designed ad hoc for the customer’s problems and are always equipped with software prototypes to run them and orchestrate the integration with public software. All results of each pipeline step are biologically interpreted at COSBI.
Continuous interaction with customers in each step of the analysis ensures timely and valuable results.
COSBI has developed modeling languages and simulation algorithms that currently set the state of the art worldwide.
COSBI is not for-profit, but it is completely self-funded. Income from services covers salaries, IT infrastructure and minimal overhead.
Complete confidentiality and data security is ensured. State of the art security IT infrastructure and protocolsare adopted and only COSBI researchers that run the analyses can access thecustomer’s data.
COSBI integrates multiomics,multilevel and clinical data setswith diet, lifestyle and scientificliterature.
12 • COSBI - April 2015
2 network selection
1 biomarkers
INPUT
phenotype, omics andclinical data, the biomarkers from step 1 for driving the selection of the backgroundbiological network
INPUT
transcriptomics, proteomics orany data set that can be ordered or measured and the phenotype of interest
COSBI defines the backgroundbiological network from the phenotype description andthe raw data provided by thecustomer.
COSBI performs the analysisstarting from raw data andphenotype description. Theresult is the set of patients’signature, the similaritynetwork and the consensussignature per luster ofpatients.
METHOD
the phenotype and the data available determine the components of the biological process of interest (genes,proteins, metabolites, drugs, nutrients, lipids, drugs), their localization, the reference tissues and organs. An ad hocnetwork is built by integrating the data provided by the customer and public knowledge from DBs and literature
METHOD
data are ordered for each patient and a signature is made up to the upper and lower elements of the ordered list (e.g. most and least expressed genes). A distance is definedbetween each pair of signatures and it is visualized on a network of patients with the length of the arcs proportional to the distance of the signatures. The closer the patients, the more similar, andthe visualization produces clusters of individuals (e.g.,health vs disease, responder vs non responder). Classical methods are applied for comparison
OUTCOME
a biological network withhighlighted biomarkers (genes, proteins, etc.) and the experimental data (e.g., mostand least expressed genes). A report with the literature references, DBs and methodology adopted is always provided
OUTCOME
a network of patientsclustered according to thephenotype. New patients can be mapped onto the networkaccording to their signatureto determine the cluster theyenter in. The correspondingstratification of patients can beused for diagnosis or prognosis, optimal selectionof cohorts for clinical trials,toxicology studies, etc.
-4 -2 0 2 4
-4-2
02
4
LD1
LD
2
glucoseIntolerant
normal
glucoseIntolerant
normal
diabetic
normal
diabetic
diabeticdiabetic
glucoseIntolerant
normal
glucoseIntolerant
diabetic
glucoseIntolerantglucoseIntolerant
normal
diabetic
diabetic
normal
diabetic
normal
normal
diabetic
normal
diabetic
glucoseIntolerant
diabetic
diabetic
normalnormal
diabetic
diabetic
glucoseIntolerant
diabetic
glucoseIntolerant
normal
normal
diabetic
diabetic
normal
diabetic
glucoseIntolerant
normal
diabeticdiabetic
normal
normal
glucoseIntolerant
diabetic
normal
diabetic
diabeticdiabetic
glucoseIntolerant
glucoseIntolerant
normal
diabetic
diabeticnormal
diabetic
diabeticdiabeticdiabetic
diabetic
glucoseIntolerant
normal
glucoseIntolerant
diabetic
diabetic
diabetic
glucoseIntolerant
diabetic
normal
diabeticdiabetic
diabeticdiabetic
normaldiabetic
normal
normal
normal
glucoseIntolerant
normal
glucoseIntolerant
normal
glucoseIntolerant
normal
diabetic
normal
normalnormal
normal
glucoseIntolerant
normal
normal
diabetic
normalnormal
normal
normal
glucoseIntolerant
normal
glucoseIntolerant
diabetic
normal
diabetic
normal
glucoseIntolerant
glucoseIntolerant
normal
normalnormal
glucoseIntolerant
diabetic
glucoseIntolerant
normal
normal
-25 -20 -15 -10 -5 0 5 10
-10
-50
5
PC1
PC
2
glucoseIntolerant
normal
glucoseIntolerant
normal
diabetic
normal
diabetic
diabetic
diabetic
glucoseIntolerant
normalglucoseIntolerant
diabeticglucoseIntolerant
glucoseIntolerantnormal
diabetic
diabetic
normal
diabetic
normal
normaldiabetic
normal
diabetic
glucoseIntolerant
diabeticdiabetic
normal
normal
diabetic
diabetic
glucoseIntolerant
diabetic
glucoseIntolerant
normalnormaldiabetic
diabetic
normal
diabetic
glucoseIntolerant
normal
diabetic
diabeticnormal
normal
glucoseIntolerant
diabetic
normal
diabetic
diabeticdiabetic
glucoseIntolerant
glucoseIntolerantnormal
diabetic
diabetic
normal
diabeticdiabetic
diabetic
diabetic
diabetic
glucoseIntolerant
normalglucoseIntolerant
diabetic
diabetic
diabetic
glucoseIntolerant
diabeticnormal
diabeticdiabetic
diabetic
diabetic
normal
diabetic
normalnormal
normalglucoseIntolerantnormal glucoseIntolerant
normal
glucoseIntolerantnormal
diabetic
normal
normal
normalnormalglucoseIntolerant
normal
normal
diabetic
normal
normalnormal
normal
glucoseIntolerant
normal
glucoseIntolerant
diabetic
normal
diabetic
normal
glucoseIntolerant
glucoseIntolerant
normal
normal
normal
glucoseIntolerant
diabetic
glucoseIntolerant
normalnormal
(A) (B)
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 13
3 network analysis
INPUT
background biological network from step 2 and available experimental data
COSBI defines the mostsignificant network modules for the phenotype and produces a report of the biological interpretation with new hypotheses for new experiments
METHOD
the biological network is studied according to topological indexes and the mapping of the experimentaldata to determine the importance of the nodes.Enrichment analysis with respect to the data available is used to define the modules. nformation theory is applied toinvestigate the transmission capacity of cascades andscore them accordingly to select the dominant pathways
OUTCOME
a set of modules (usually different from classicalpathways as they aredetermined by experimentaldata) and measures of statistical significance, activity level, sensitivity, robustness. The most topologically relevant nodes for each module are identified with respect to the phenotype of interest
4 functional analysis
INPUT
network modules from step 3 and phenotype of interest
COSBI produces a biologicalinterpretation of the modules and suggests possible targetsor modulators of the biologicalprocesses and new experiments to elucidate molecular mechanisms
METHOD
modules from step 3 are tuned to the phenotype via enrichment analysis with respect to the functions of their components. Literature mining is applied to compare our analysis results with the biological knowledge in the literature and DBs. The primarymetabolic and signaling cascades are identified andassociated with biological functions
OUTCOME
a report with metrics of the module for the statistical significance, coverage of known pathways and biological interpretation of the enrichment analysis. The modules are the parts of the network where to look fortargets and modulators or where to investigate to better understand molecular mechanisms
NOTCH1 Intracellular Domain Regulates TranscriptionSignaling by NOTCH1 PEST Domain Mutants in Cancer
Signaling by NOTCH1 HD+PEST Domain Mutants in CancerSignaling by NOTCH1 in Cancer
Signaling by NOTCH1 HD Domain Mutants in CancerFBXW7 Mutants and NOTCH1 in Cancer
Signaling by NOTCH1Signaling by NOTCH1 t(79)(NOTCH1:M1580_K2555) Translocation Mutant
Constitutive Signaling by NOTCH1 PEST Domain MutantsConstitutive Signaling by NOTCH1 HD+PEST Domain MutantsTranscriptional Regulation of White Adipocyte Di erentiation
Fatty acid, triacylglycerol, and ketone body metabolismRegulation of Lipid Metabolism by Peroxisome proliferator-activated receptor alpha
PPARA Activates Gene ExpressionRORA Activates Circadian Expression
Circadian Repression of Expression by REV-ERBACircadian Clock
BMAL1:CLOCK/NPAS2 Activates Circadian ExpressionYAP1- and WWTR1 (TAZ)-stimulated gene expression
Regulation of Cholesterol Biosynthesis by SREBP (SREBF)Activation of Gene Expression by SREBP (SREBF)
Metabolism of lipids and lipoproteinsDevelopmental Biology
Generic Transcription PathwayNuclear Receptor transcription pathway
Bile acid and bile salt metabolismalpha-linolenic acid (ALA) metabolism
alpha-linolenic (omega3) and linoleic (omega6) acid metabolismCD28 dependent PI3K/Akt signaling
Release of eIF4ES6K1-mediated signalling
mTORC1-mediated signallingRecycling of eIF2:GDP
LDL-mediated lipid transportScavenging by Class B Receptors
HDL-mediated lipid transportRetinoid metabolism and transport
Chylomicron-mediated lipid transportLipid digestion, mobilization, and transport
Lipoprotein metabolismTriglyceride Biosynthesis
Synthesis of very long-chain fatty acyl-CoAsFatty Acyl-CoA Biosynthesis
Cell Cycle, MitoticCell Cycle
G2/M TransitionMitotic G2-G2/M phases
Regulation of PLK1 Activity at G2/M TransitionRecruitment of mitotic centrosome proteins and complexes
Centrosome maturationLoss of Nlp from mitotic centrosomes
AMPK inhibits chREBP transcriptional activation activityImport of palmitoyl-CoA into the mitochondrial matrix
Signaling by Insulin receptorPI3K Cascade
IRS-related events triggered by IGF1RIGF1R signaling cascade
Signaling by Type 1 Insulin-like Growth Factor 1 Receptor (IGF1R)Insulin receptor signalling cascade
IRS-mediated signallingIRS-related events
Regulation of Rheb GTPase activity by AMPKPKB-mediated events
mTOR signallingRegulation of AMPK activity via LKB1
Energy dependent regulation of mTOR by LKB1-AMPK
! "! #! $!
Loss of proteins required for interphase microtubule organization
number of pathway genes in module
25
50
75
100
% pathwaycoverage
14 • COSBI - April 2015
5 simulation
INPUT
relevant modules for the phenotype from step 4 and hypothesis to be verified
COSBI produces the executable model and runsthe simulation corresponding to the virtual experiment
METHOD
modules are represented in a graphical language suitable to interact with biologists and to simulate (either deterministically or stochastically) the dynamics of the biological process. After parameter inference and model calibration, perturbation experiments are performed in silico to verify the hypotheses
DATA TYPES
OMICS (all platforms)genetics, SNPs, genomics, proteomics, transcriptomics, metabolomics, lipidomics, microbiome
CLINICALblood markers, urine and saliva markers, tissue markers, diet, physical activity
BIOCHEMICALkinetic parameters and reaction rates, affinity, active domains and binding sites
LITERATUREmining and search to acquire the knowledge needed to optimize the pipeline and drive the analysis especially in the step of network selection and to assess the results
DATA BASEScomparison, integration and validation of the analyses through public data collections
METHODS
DATA ANALYSISstandard and multivariate statistics, data exploration, aggregation and visualization, clustering, functional and topological network analysis, dimensionality reduction, annotation, data and literature mining, machine learning, proprietary methods defined ad hoc for specific biological problems, sensitivity and robustness analysis
MODEL REPRESENTATIONpublic and proprietary graphical languages, domain-specific and general purpose programming languages, reaction-based modeling, agent-based modeling, differential equations, boolean networks, Petri nets, rewriting systems
SIMULATION ALGORITHMSdeterministic, stochastic and hybrid algorithms, non parametric simulation, incomplete model simulation, proprietary algorithms to identify dominant pathways
OUTCOME
executable model of thebiological processes ofinterest and biological interpretation of the virtual experiments; preliminary validation through literature search.Elucidation of the mechanisms of action, design of new experiments
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 15
Main studies
Cofactor network. Integration of data from multiple sources in order to build a comprehensive network linking cofactors, cofactor-requiring enzymes, human diseases and biological processes. Study of the influence ofbetween-population genetic differences on polymorphic distribution in cofactor-requiring genes
Assessment of CVD risk. Identification of healthy individuals at low (controls) and high-CV (cases) risk based on fasting proteomic signature of data and of the genetic factors involved in ischemic stroke predisposition
Modeling the response of small intestine to dietary fat intake. Identification of genes exhibiting significant linear or nonlinear response to dietary fat doses. Identification of dietary fat responsive metabolic and transport processes that are commonly enriched from the proximal to distal sections; commonly affected intestinal segments; Predominant transcriptomic response patterns
Compound-affected differentiation. Cell differentiation and comparative differences among different dosages of compounds over multiple time points from metabolomics andproteomics
Methods
data and literature mining,network analysis (moduleidentification, centrality, hubidentification, dominator tree),permutation test, functionalenrichment analysis andbiological interpretation, FSTindex, data visualization
functional enrichment analysis and biological interpretation, clustering, rank-based signatures, network analysis, genetic algorithms, PCA, casecontrol association analyses (chi-square and logistic regression) using genomewide SNP, data visualization
nonlinear regression of transcriptomic data, weighted co-expression network analysis, hypergeometric test of functional enrichmentanalysis, biologicalinterpretation
integrative multiomic analysis,statistical indices, normalization and variance stabilization, differential analyses, co-expression analysis, data mining, network analysis, dominant pathway identification, functional enrichment analysis and biological interpretation, data visualization
man/ months*
6-10
10-14
10-14
10-14
16 • COSBI - April 2015
Main studies
Response to metabolic challenges. Identification of proteins and pathways that are differentially expressed after a metabolic challenge. Identification of genetic and proteomic markers predictive of metabolic challenge response and development of a systems view of these responses
Metabolic flexibility and metabolic phenotypes. Different measures of fat and carbohydrate oxidation, and individual levels of severalinflammatory markers in lipid and glucose oral tolerance tests (OLTT and OGTT). Investigation on gene expression patterns in OLTT and OGTTand inflammatory patterns. Investigation of the effect of diet to predict of the levels of someinflammatory markers
SNPs predicting diabetes phenotype. Investigation on groups of SNPs to see if they have a predictive role on several diabetes related phenotypes on a wide (~800) cohort of diabetic subjects
Extreme phenotypes of diabetic patients. Gene expression microarray data analysis in extreme phenotypes of 148 diabetic subjects (10 most insulin resistant and 10 most insulin sensitive)
Systemic response to food intake of T1 diabetic patients. Development of a mathematical model of mixed meal in type 1 diabetes
Methods
integrative multiomic analysiscovariate analysis, robustlinear regression, multiplecorrection testing, functionalenrichment analysis andbiological interpretation, dataand literature mining, GWAS,pQTL, robust sparse k-meansclustering, rank-basedsignatures, genetic algorithms,data visualization
integrative multiomic analysis ,GWAS, pathway identificationand biological interpretation,data visualization, multipleregression models, canonicalcorrelation analysis, mixomics,PMA, proprietary algorithms,gene enrichment analysis,network analysis
normalization, random forest,data and literature mining,biological interpretation, dataexploration and visualization
differential analysis, probemapping, gene enrichmentanalysis and interpretation,rank based signature, geneticalgorithms, data and literaturemining, data visualization
data and literature mining,ODE, simulation algorithms,virtual experiments, datavisualization and biologicalinterpretation
man/ months*
8-12
10-14
3-6
3-6
8-12
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 17
Main studies
Role of leptin in food intake and energy metabolism. Development of a mathematical model of leptin dynamics with parameters derived from published experimental data
Sphingolipid metabolism. Network selection, module identification and development of a mathematical model of sphingolipid dynamics from literature knowledge and experimental data
EGFR internalization. Development of a mathematical model of the molecular mechanisms driving internalization of the EGF receptor in HeLa cells from literature and experimental data
Inference of cancer gene essentiality from genomic data. Characterization of cancer cell lines with biomarkers for tailored treatments and patients with higher treatment efficacy
Neurodegenerative dementia. Investigation of the molecular connections between complexdiseases with the shared clinical symptoms of dementia
Methods
data and literature mining,ODE, simulation algorithms,virtual experiments, datavisualization and biologicalinterpretation
data and literature mining,interaction network selection,differential analysis,enrichment analysis,functional analysis, moduleidentification, ODE and ad hocprogramming languages,simulation algorithms, virtualexperiments, datavisualization and biologicalinterpretation
data and literature mining,ODE and stochastic models,proprietary language,simulation algorithms, virtualexperiments, fitting, modelcalibration
integrative multiomic analysis,support vector regression (e-SVR), clustering, data andliterature mining, differentialanalysis, network analysis,biological interpretation, datavisualization
multi source data integration,literature and data mining,machine learning, networkanalysis, rank-basedsignatures, genetic algorithms,biological interpretation, datavisualization
man/ months*
8-12
10-14
10-14
4-6
6-8
18 • COSBI - April 2015
Main studies
Novel dementia drug targets. A multi-relational association mining approach to predict targets for innovative therapeutic treatment ofdementia
Alzheimer’s and diabetes crossdiseaseanalysis. Integration of transcriptomic data from Alzheimer’s and diabetes post mortem brains toassess commonalities between two highly co-morbid diseases
Role of APOE4 in Alzheimer’s.Analysis of the role of one of the genetic predisposing genetic factors (APOE4) of Alzheimer’s
A multi-factor network analysis onAlzheimer’s. Integration of the genomic aspect of AD with the gene expression and drug candidate targets for the understanding of disease pathophysiology
Neurological diagnostic biomarkers.Early diagnosis of neurological disorders (autism, Parkinson’s) using gene expression profiles
Toxicology. Diagnosis of hepatotoxicity using gene expression profiles
Methods
multi relational mining, dataand literature mining, dataintegration and exploration,statistical indices, networkanalysis, biologicalinterpretation, datavisualization
differential analysis, data andliterature mining, networkanalysis, rank-based signature,enrichment analysis,functional analysis, moduleidentification, datavisualization and biologicalinterpretation
integrative multiomic analysis,data and literature mining,network analysis, rank-basedsignature, enrichment analysis,functional analysis, datavisualization and biologicalinterpretation
multi source data integration,data and literature mining,network analysis, enrichmentanalysis and biologicalinterpretation, datavisualization, machine learning
data normalization andfiltering, rank-based signature,genetic algorithms, data andliterature mining, pathwayanalysis and interpretation
data normalization andfiltering, rank-based signature,genetic algorithms, pathwayanalysis and interpretation
man/ months*
6-8
8-12
8-10
4-6
4-6
4-6
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 19
the Delta Obesity Vitamin Study. Genes & Nutrition, 9(3):403, 2014.15. J. Kaput, B. Van Ommen, B. Kremer, C. Priami, J. Pontes Monteiro, M.J. Morine,
et al. Consensus statement understanding health and malnutrition through a systems approach: the ENOUGH program for early life. Genes & Nutrition, 9:1-9, 2014.
16. M.J. Morine, J. Pontes Monteiro, C. Wise, C. Teitel, L. Pence, A. Williams, B. Ning, B. McCabe-Sellers, C. Champagne, J. Turner, B. Shelby, M. Bogle, R.D. Beger, C. Priami, J. Kaput. Genetic associations with micronutrient levels identified in immune and gastrointestinal networks. Genes & Nutrition, 9(4):408, 2013.
17. R. De Cegli, S. Iacobacci, G. Flore, G. Gambardella, L. Mao, L. Cutillo, M. Lauria, J. Klose, E. Illingworth, S. Banfi, D. di Bernardo. Reverse engineering a mouse embryonic stem cell-specific transcriptional network reveals a new modulator of neuronal differentiation. Nucleic Acids Res, 41:711-26, 2013.
18. M. Scotti, L. Stella, E. Shearer, P. Stover. Modeling cellular compartmentation in one- carbon metabolism. WIREs Systems Biology and Medicine, 5:343-365, 2013.
19. 13. J. Dodgson, A. Chessel, M. Yamamoto, F. Vaggi, S. Cox, E. Rosten, D. Albrecht, M. Geymonat, A. Csikasz-Nagy, M. Sato, R. E. Carazo-Salas. Spatial segregation of polarity factors into distinct cortical clusters is required for cell polarity control. Nature Communications, 2013.
20. L. Caberlotto, M. Lauria, P. Nguyen, M. Scotti. The central role of AMP-kinase and energy homeostasis impairment in Alzheimer’s disease: a multifactor network analysis. PloS ONE, 8:e78919, 2013.
21. A. Tarca, M. Lauria, M. Unger, E. Bilal, S. Boue, K. Dey, J. Hoeng, H. Koeppl, F. Martin, P. Meyer, P. Nandy, R. Norel, M. Peitsch, J. Rice, R. Romero, G. Stolovitzky, M. Talikka, Y. Xiang, C. Zechne. Strengths and limitations of microarray-based phenotype prediction: Lessons learned from the IMPROVER Diagnostic Signature Challenge. Bioinformatics, 29:2892-9, 2013.
22. M. Lauria. Rank-based transcriptional signatures: A novel approach to diagnostic biomarker definition and analysis. Systems Biomedicine, 1:0-10, 2013.
23. O. Kahramanogullari, J. Lynch. Stochastic Flux Analysis of Chemical Reaction Networks. BMC Systems Biology, 7:133, 2013.
24. K. Martin, M.J. Morine, J. Hager, B. Sonderegger, J. Kaput. Perspective: a systems approach to diabetes research. Front Genet, 4:205, 2013.
25. C. O’Grada, M.J. Morine, C. Morris, M. Ryan, E. Dillon, M. Walsh, E. Gibney, L. Brennan, M. Gibney, H. Roche. PBMCs reflect the immune component of the WAT transcriptome - Implications as biomarkers of metabolic health in the postprandial state. Molecular nutrition & food research, 2013.
26. J. Kaput, M.J. Morine. Discovery-Based Nutritional Systems Biology: Developing N-of-1 Nutrigenomic Research. Int J Vitam Nutr Res, 82(5):333-41, 2012.
1. M. Lauria, P. Moyseos, C. Priami. SCUDO: a tool for signature-based clustering of expression profiles. Nucleic Acid Research, 2015.
2. H. Vo Thanh, R. Zunino, C. Priami. On the Rejection-based Algorithm for Simulation and Analysis of Large-Scale Reaction Networks. Journal of Chemical Physics, 142:244106, 2015.
3. H. Vo Thanh, C. Priami. Simulation of Biochemical Reactions with Time- Dependent Rates by the Rejection-based Algorithm. Journal of Chemical Physics, 143, 2015.
4. T.-P. Nguyen, C. Priami, L. Caberlotto. Novel Drug Target Identification for the Treatment of Dementia Using Multi-Relational Association Mining. Nat. Sci. Rep., 5:11104, 2015.
5. F. Capuani, A. Conte, E. Argenzio, L. Marchetti, C. Priami, S. Polo, P.P. Di Fiore, S. Sigismund, A. Ciliberto. Quantitative analysis reveals how EGFR activation and downregulation are coupled in normal but not in cancer cells. Nat. Comm., 2015.
6. S. Rizzetto, C. Priami, A. Csikász-Nagy. Qualitative and Quantitative Protein Complex Prediction Through Proteome-Wide Simulations. PLOS Comp. Biol., 2015.
7. O. Finucane, C. Lyons, A. Murphy, C. Reynolds, R. Klinger, N. Healy, A. Cooke, R. Coll, L. McAllan, K. Nilaweera, M. O’Reilly, A. Tierney, M.J. Morine, J. Alcala-Diaz, J. Lopez-Miranda, D. O’Connor, L. O’ Neill, F. McGillicuddy, H. Roche. Monounsaturated fatty acid enriched high fat-diets impede adipose NLRP3 inflammasome mediated IL-1β secretion and insulin resistance despite obesity. Diabetes, 2015.
8. P. Nguyen, L. Caberlotto, M.J. Morine, C. Priami. Network Analysis of Neurodegenerative Disease Highlights a Role of Toll-Like Receptor Signaling. BioMed Research International, 2014:1-16, 2014.
9. L. Caberlotto, P. Nguyen. A Systems Biology investigation of Neurodegenerative Dementia reveals a pivotal role of Autophagy, BMC Systems Biology, 8:65, 2014.
10. L. Caberlotto, M. Lauria. Systems biology meets -omic technologies: novel approaches to biomarker discovery and companion diagnostic development. Expert Review of Molecular Diagnostics, November:1-11, 2014.
11. M. Lauria. Rank-Based miRNA Signatures for Early Cancer Detection. BioMed Research International, Vol. 2014:Article ID 192646, 2014.
12. R. Gostner, B. Baldacci, M.J. Morine, C. Priami. Graphical Modeling Tools for Systems Biology. ACM Computing Surveys, 47(2), 2014.
13. T. Vo, C. Priami, R. Zunino. Efficient Rejection-based Simulation of Biochemical Reactions with Stochastic Noise and Delays. J. Chem. Phys., 141, 2014.
14. J. Pontes Monteiro, C. Wise, M.J. Morine, C. Teitel, L. Pence, A. Williams, B. McCabe-Sellers, C. Champagne, J. Turner, B. Shelby, B. Ning, J. Oguntimein, L.Taylor, T. Toennessen, C. Priami, R.D. Beger, M. Bogle, J. Kaput. Methylation potential associated with diet, genotype, protein, and metabolite levels in
The main methods developed at COSBI in ten years of activities and tested both on industrial projects and classroom at the Uni-versity of Trento are introduced in the book “analysis of biologicalsystems,” Imperial College Press, March 2015.
Imperial College PressImperial College PressP1004 hc
www.icpress.co.uk
ISBN 978-1-78326-687-6
corrado pr iami mel issa j . morine
analys is of biological systems
Modeling is fast becoming fundamental to understanding the processes that define biological systems. High-throughput technologies are producing increasing quantities of data that require an ever-expanding toolset for their e�ective analysis and interpretation. Analysis of high-throughput data in the context of a molecular interaction network is particularly informative as it has the potential to reveal the most relevant network modules with respect to a phenotype or biological process of interest.
Analysis of Biological Systems collects classical material on analysis, modeling and simulation, thereby acting as a unique point of reference. The joint application of statistical techniques to extract knowledge from big data and map it into mechanistic models is a current challenge of the field, and the reader will learn how to build and use models even if they have no computing or math background. An in-depth analysis of the currently available technologies, and a comparison between them, is also included. Unlike other reference books, this in-depth analysis is extended even to the field of language-based modeling. The overall result is an indispensable, self-contained and systematic approach to a rapidly expanding field of science.
analysis of biological systems
priamimorine
analys is of b iological systems
Main recent publications
COSBI Methods Book
20 • COSBI - April 2015
27. F. Niola, X. Zhao, D. Singh, A. Castano, R. Sullivan, M. Lauria, H. Nam, Y. Zhuang, R. Benezra, D. di Bernardo, A. Iavarone, A. Lasorella. Id proteins synchronize stemness and anchorage to the niche of neural stem cells. Nature Cell Biology, 14:477-8, 2012.
28. P. Lecca, D. Morpurgo, G. Fantaccini, A. Casagrande, C. Priami. Inferring biochemical reaction pathways: the case of the gemcitabine pharmacokinetics, BMC Systems Biology, 6:51, 2012.
29. O. Kahramanogullari, G. Fantaccini, P. Lecca, D. Morpurgo, C. Priami. Algorithmic Modeling Quantifies the Complementary Contribution of Metabolic Inhibitions to gemcitabine Efficacy. PLoS ONE, 7:12, 2012.
30. P. Nguyen, T. Ho. Detecting Disease Genes Based on Semi-Supervised Learning and Protein-Protein Interaction Networks. Artificial Intelligence in Medicine, 54. 1:63-71, 2012.
31. A. Romanel, L. Jensen, L. Cardelli, A. Csikasz-Nagy. Transcriptional Regulation Is a Major Controller of Cell Cycle Transition Dynamics. PLoS ONE, 7:e29716, 2012.
32. S. Lai, W. Liu, F. Jordan. On the centrality and uniqueness of species from the network perspective. Biology Letters, 8:570-573, 2012.
33. F. Jordan, P. Nguyen, W. Liu. Studying protein-protein interaction networks: a systems view on disease. Briefings in Functional Genomics, 2012.
34. F. Ferrezuelo, N. Colomina, A. Palmisano, E. Garí, C. Gallego, A. Csikasz-Nagy, M. Aldea. The critical size is set at a single-cell level by growth rate to attain homeostasis and adaptation. Nature Communications, 3:1012, 2012.
35. L. Cardelli, A. Csikasz-Nagy. The Cell Cycle switch Computes Approximate Majority. Scientific Reports, 2:656, 2012.
36. P. Lecca, C. Priami. Biological network inference for drug discovery. Drug Discovery Today, 2012.
37. E. Brennan, M.J. Morine, D. Walsh, S. Roxburgh, M. Lindenmeyer, D. Brazil, P. Gaora, H. Roche, D. Sadlier, C. Cohen, C. Godson, F. Martin. Next-generation sequencing identifies TGF-ß1- associated gene expression profiles in renal epithelial cells reiterated in human diabetic nephropathy. Biochimica et Biophysica Acta (BBA)-Molecular Basis of Disease, 2012.
38. E. Allott, M.J. Morine, J. Lysaght, S. McGarrigle, C. Donohoe, J. Reynolds, H. Roche, G. Pidgeon. Elevated tumour expression of PAI-1 and SNAI2 in obeseoesophageal adenocarcinoma patients and impact on prognosis. Clinical and Translational Gastroenterology, 2012.
39. C. Reynolds, S. Toomey, R. McBride, J. McMonagle, M.J. Morine, O. Belton, A. Moloney, H. Roche. Divergent effects of a CLA-enriched beef diet on metabolic health in ApoE(-/-) and ob/ob mice. Journal of Nutritional Biochemistry, 2012.
40. M.J. Morine, S. Toomey, F. McGillicuddy, C. Reynolds, K. Power, J. Browne, C. Loscher, K. Mills, H. Roche. Network analysis of adipose tissue gene expression highlights altered metabolic and regulatory transcriptomic activity in high-fat-diet-fed IL-1RI knockout mice. Journal of Nutritional Biochemistry, 2012.
41. N. Gjata , M. Scotti, F. Jordan. The strength of simulated indirect interaction modules in a real food web. Ecological Complexity, 11:160-164, 2012.
42. M. Scotti, N. Gjata , C. Livi, F. Jordan. Dynamical effects of weak trophic interactions in a stochastic food web simulation. Community Ecology, 13:230-237, 2012.
43. F. Jordan, N. Gjata , M. Shu, C. Yule. Simulating food web dynamics along a gradient: quantifying human influence. PLoS ONE, 7:e40280, 2012.
2014.14. Visual Modeling of Biological Systems. Fraunhofer Institute for Algorithms
and Scientific Computing, Sankt Augustin, Germany, Jun 2014.15. A novel approach to systems pharmacology. Drug Discovery Summit,
Geneva, Jun 2014.16. Algorithms for ecological network analysis. 6th SIDEER Symposium, Sede
Boqer Campus of Ben Gurion University, Israel, Mar 201417. Systems biology: a molecular nutrition perspective. Molecular-Med Tri-Con,
San Francisco, Feb 2014.18. Key species and key interactions in food web simulations. University of
Potsdam, Potsdam, Germany, Jan 2014.
19. Key players in ecological networks. Food Webs 2013 Symposium, Giessen, Germany, Nov 2013.
20. Network dynamics: from data to behavior. Merck, New York, Oct 2013.21. Network science as the key for understanding complex problems at different
spatial scales. Academia Sinica, Taipei, Taiwan, Sep 2013.22. Dynamic simulation of Biological Systems. Sanofi, Frankfurt, Feb 2013.23. Bioinformatics at COSBI, Microsoft Italy, Feb 2013.24. Network identification, analysis and simulation, Amgen, Los Angeles, Feb
2013.25. Systems biology: a molecular nutrition perspective. Molecular-Med Tri-Con,
San Francisco, Feb 2013.26. Service-oriented data aggregation, analytics and interactive visualization.
Google, Mountain-View, Jan 2013.
1. Key players in the microbial ecosystems of the human body, Discovery on Target, Boston, Sep 2015.
2. Sphingolipid metabolism and data-driven module detection, Sanofi, Frankfurt, Jul 2015.
3. Computational systems biology applied to pharmacology and nutrition, ISC, Frankfurt, Jul 2015.
4. Systems level understanding of biological processes in nutrition, Nestlé Institute of Health Science, Lausanne, Jul 2015.
5. Ranking omics data for discovering biomarkers, Mol-Med Tri-Conference, San Francisco 1, Feb 2015.
6. The COSBI case. Big Data Leaders Forum, Berlin, Dec 2014.7. Quantitative analysis of biological systems. Energy Biosciences Institute,
University of Berkeley, Sep 2014.8. bSTYLE - a minimal graphical language to model biological systems.
Microsoft, Redmond, Sep 2014.9. Quantitative network analysis of biological systems. SomaLogic, Boulder, Sep
2014.10. Programming languages and biology. Microsoft Research Cambridge, Sep
2014.11. Quantitative pipelines for systems pharmacology. GSK, Stevenage, Sep 2014.12. Identification of cofactor-requiring enzymes with high genetic differentiation
between 1000 Genomes populations. European NuGO week, Castellammare di Stabia, Sep 2014.
13. Are Alzheimer’s disease and neurodegenerative dementia primarily metabolic diseases? A systems biology study. University of Bologna, Jun
Main recent invited talks
The Microsoft Research - University of Trento Centre for Computational and Systems Biology • 21
Main competitive grants
Where we are
1. MIUR-FIRB, Computational tools for systems biology, 2005-2009.2. PAT, Language-based systems biology, 2006-2009.3. CARITRO, Molecular modelling of polyphenols biosynthetic pathways and its application in gene discovery and gene modulation in living cells for pharmacogenomics purposes, 2007-2009.4. CARIPLO NOBEL, Molecular modeling of gene regulation, transcription and translation, 2007-2011.5. HFSP, Quantitative study of polarised cell growth in vivo and in silico, 2009-2011.6. PAT, Personalized molecular nutrition, 2010-2014.7. EU (Action: Integrated Infrastructure Initiative (I3)) VENUS-C: Virtual multidisciplinary EnviroNments USing Cloud infrastructures, 2010-2012.8. EU (JPI - Infrastructures), ENPADASI - A healthy diet for healthy life. 2014-2016.
Dissemination is fundamental for COSBI to keep its strategic partnerships alive and toincrease its visibility in the scientific and industrial communities.
27. Algorithmic Systems Biology: from omics data set to mechanistic models. HITS, Germany, Jan 2013.
28. Computational and Systems Biology at COSBI. UCB, Brussels, Jan 2013.
29. Algorithmic Systems Biology: from omics data set to mechanistic models, FOSBE 2012, Tokyo, Japan, Oct 2012.
30. Mastering complexity of biological systems through network modularization. BIO-IT World Europe, Vienna, Oct 2012.
31. Computing as Enabling Technology for Systems Biology. SEFM, Greece, Sep 2012.
32. High-throughput analysis of nutritional heath, Atlantic Food and Horticultural Research Centre. Kentville, Canada, Jul 2012.
33. Food web dynamical simulations. FiBL, Frick, Switzerland, Mar 2012.34. Algorithmic systems biology: mastering the complexity of biosystems
without math and computing background. Applications of Systems Biology in Drug Discovery and Development Mini Symposium, Basel and International Conference and Exhibition on Biometrics & Biostatistics, Omaha, USA Mar 2012; Molecular Med Tri-Con, San Francisco, and Pharmaceutica, San Francisco Feb 2012.
>450COSBI sc
ientific papers
1 paper every 6 days
>5500COSBI citations
>1 citation per day 22COSBI innovative S W
>2 per year
scientific events
organized by COSBI26~3 per year
seminarsorganized by COSBI
2891 every 18 days
COSB
I invit
ed to scientific events
>260
1 invit
ation
every 2 weeks
>30 videos
COSBI invited talks
~210 >200h presentations>7500 slides
1 every 18 days
events organized by COSBI5 student competitions
8 school visits
non scientific dissemination
~3 per year, >800 participants
27 8 artistic exhibitions
6 open doors and dissemination
23 stages, 60 in events
16 bachelor theses
32 master theses
6 international master theses
16 PhD theses
1 student every 7 days contacts COSBI
COSBI in education: COSBI theses70
COSBI media clips
1 clip every 2 week 640 24 radio clips 585 press clips
31 TV clips
4,3avg IF
>165% avg world fieldCOSBI research quality
>16,5M>1,6 M per year
COSBI non local funds
COSBI in presti
gious
scientific bodies world
-wide
17
~1 per resea
rcher
COSBI educated people
in worldwide research13 in local research system
34
www.cosbi.eu