Network medicine
Applicazioni:
L’impatto dei network
cellulari sulla comorbidity
Studio della comorbidity
Le interazioni tra molecole a livello cellulare
correlano a livello di popolazione con
i pattern di comorbidity?
L impatto dei network cellulari sulla comorbidity
Morbidity (from Latin morbidus, meaning "sick, unhealthy")
= is a diseased state, disability, or poor health due to any cause.
= incidence of a particular disease in a population
Morbilità/Morbosità indica uno stato patologico dell'individuo
Comorbidity = comorbosità, cioè coesistenza di stati
patologici/malattie nella stessa persona.
Comorbility = comorbilità, che, in ambito statistico, indica
incidenza o prevalenza di più patologie in una particolare
area geografica.
Comorbidity - considerazioni generali e domande aperte
1. Le funzioni cellulari sono controllate da reti di geni, proteine,
metaboliti, etc. Diffetti in un modulo si possono propagare ad altri moduli?
2. Le interdipendenze tra i network/subnetwork cellulari sono più importanti
a livello di individuo oppure anche di popolazioni?
3. La comorbidity è influenzata dallo stile di vita, condizioni ambientali,
e terapie.
4. Le malattie presenti nel network delle malattie (studio sul diseasoma)
presentano una comorbidity significativa?
5. Le malattie genetiche che condividono geni e le proteine codificate da
questi geni interagiscono, presentano comorbidity?
6. Le malattie che hanno un alto livello di geni coespressi presentano
comorbidity?
Le tappe dello studio della comorbidity delle
malattie genetiche
1. Costruzione della rete di malattie genetiche
2. Analisi della comorbidity
3. Predizioni / Conclusioni
Goh K et al. PNAS 2007;104:8685-8690
2. Analisi della comorbidity
Classificazione delle malattie
"The International Classification of Diseases, 9th Revision, Clinical
Modification" (ICD-9-CM), issued for use beginning October 1, 2008 for federal
fiscal year 2009 (FY09). http://www.cms.gov/
http://www.cms.gov/MedicareGenInfo/
http://www.medicare.gov/
The ICD-9-CM is maintained jointly by the National Center for Health Statistics
(NCHS) and the Centers for Medicare & Medicaid Services (CMS).
http://www.icd9data.com/
001-139
140-239
240-279
280-289
290-319
320-389
390-459
460-519
520-579
580-629
630-679
680-709
710-739
740-759
760-779
780-799
800-999
INFECTIOUS AND PARASITIC DISEASES
NEOPLASMS
ENDOCRINE, NUTRITIONAL AND METABOLIC DISEASES, AND IMMUNITY DISORDERS
DISEASES OF THE BLOOD AND BLOOD-FORMING ORGANS
MENTAL DISORDERS
DISEASES OF THE NERVOUS SYSTEM AND SENSE ORGANS
DISEASES OF THE CIRCULATORY SYSTEM
DISEASES OF THE RESPIRATORY SYSTEM
DISEASES OF THE DIGESTIVE SYSTEM
DISEASES OF THE GENITOURINARY SYSTEM
COMPLICATIONS OF PREGNANCY, CHILDBIRTH, AND THE PUERPERIUM
DISEASES OF THE SKIN AND SUBCUTANEOUS TISSUE
DISEASES OF THE MUSCULOSKELETAL SYSTEM AND CONNECTIVE TISSUE
CONGENTIAL ANOMALIES
CERTAIN CONDITIONS ORIGINATING IN THE PERINATAL PERIOD
SYMPTOMS, SIGNS, AND ILL-DEFINED CONDITIONS
INJURY AND POISONINGRY AND POISONING
http://www.icd9data.com/2011/Volume1/140-239/170-175/170/default.htm
2011 ICD-9-CM Diagnosis Code 170
Malignant neoplasm of bone and articular cartilage
2011 ICD-9-CM Diagnosis Code 170.0
Malignant neoplasm of bones of skull and face except mandible
2011 ICD-9-CM Diagnosis Code 170.1
Malignant neoplasm of mandible
2011 ICD-9-CM Diagnosis Code 170.2
Malignant neoplasm of vertebral column excluding sacrum and coccyx
2011 ICD-9-CM Diagnosis Code 170.3
Malignant neoplasm of ribs sternum and clavicle
2011 ICD-9-CM Diagnosis Code 170.4
Malignant neoplasm of scapula and long bones of upper limb
2011 ICD-9-CM Diagnosis Code 170.5
Malignant neoplasm of short bones of upper limb
2011 ICD-9-CM Diagnosis Code 170.6
Malignant neoplasm of pelvic bones sacrum and coccyx
2011 ICD-9-CM Diagnosis Code 170.7
Malignant neoplasm of long bones of lower limb
2011 ICD-9-CM Diagnosis Code 170.8
Malignant neoplasm of short bones of lower limb
2011 ICD-9-CM Diagnosis Code 170.9
Malignant neoplasm of bone and articular cartilage site unspecified
Studio della comorbidity
2. Analisi della comorbidity
Pazienti
Studio effettuato su pazienti registrati nel
database Medicare:
N= 13.039.018 pazienti registrati su Medicare per un totale di
32.341.348 visite mediche
Periodo: più di 4 anni (1990 to 1993)
Età media: 76.3 ± 7.4;
41.8% maschi; 89.9% caucasici
http://www.medicare.gov/
IL CALCOLO DELLA COMORBIDITY
i, j = malattie co-espresse
Ii = incidenza della malattia i (numero di pazienti)
Cij = numero di pazienti simultaneamente diagnosticati
con le malattie i e j
Parametri per misurare la comorbidity:
RR = Cij/Cij* = rischio relativo
Cij* = IiIj/N (valore atteso di Cij quando le due
malattie sono indipendenti)
φ correlation
RR>1
φ>0
>4.900 associazioni malattia-gene: OMIM database
(2008)
I codici ICD-9-CM (12.000) sono stati manualmente
connessi ai nomi delle malattie del database OMIM
Solamente 763 codici ICD-9-CM sono stati mappati
nel database OMIM!
90% dei pazienti Medicare
sono diagnosticati con i
codici ICD - OMIM
Esempio di due malattie che mostrano comorbidity e sono
linked a livello di network cellulare
13 PPI
Le due malattie
presentano anche
domain-sharing
nella proteina TP53
Studio della comorbidity
Tre parametri per quantificare le relazioni tra i
network cellulari e la comorbidity
Numero di disease genes condivisi tra le malattie i e j;
Quantifica la potenziale origine genetica comune delle due
malattie
Numero di PPI tra le proteine codificate dai disease genes
(non solamente dai geni condivisi) della malattia i e quelle
della malattia j; quantifica la relazione a livello di PPI delle
due malattie
Coespressione genica media; rappresenta la Pearson
correlation media della coespressione di ogni coppia di geni
di ogni malattia
La distribuzione tessuto-specifica dei disease genes
Considerazione di partenza: le proteine codificate dai disease genes
che interagiscono nello stesso modulo funzionale tendono ad essere
espresse nello stesso tessutto.
Per la co-espressione di geni si usano i dati di riferimento
pubblicati da Ge X e colleghi:
Genomics. 2005 Aug;86(2):127-41.
Interpreting expression profiles of cancers by genome-wide survey of breadth of expression in
normal tissues.
Ge X, Yamamoto S, Tsutsumi S, Midorikawa Y, Ihara S, Wang SM, Aburatani H.
Abstract
A critical and difficult part of studying cancer with DNA microarrays is data interpretation. Besides the need for data
analysis algorithms, integration of additional information about genes might be useful. We performed genomewide expression profiling of 36 types of normal human tissues and identified 2503 tissue-specific genes.
We then systematically studied the expression of these genes in cancers by reanalyzing a large collection of
published DNA microarray datasets. We observed that the expression level of liver-specific genes in hepatocellular
carcinoma (HCC) correlates with the clinically defined degree of tumor differentiation. Through unsupervised
clustering of tissue-specific genes differentially expressed in tumors, we extracted expression patterns that are
characteristic of individual cell types, uncovering differences in cell lineage among tumor subtypes. We were able to
detect the expression signature of hepatocytes in HCC, neuron cells in medulloblastoma, glia cells in glioma, basal
and luminal epithelial cells in breast tumors, and various cell types in lung cancer samples. We also demonstrated
that tissue-specific expression signatures are useful in locating the origin of metastatic tumors. Our study shows
that integration of each gene's breadth of expression (BOE) in normal tissues is important for biological
interpretation of the expression profiles of cancers in terms of tumor differentiation, cell lineage, and metastasis.
Studio della comorbidity
Le interazioni tra molecole a livello cellulare
aumentano la probabilità che gli individui
sviluppino simultaneamente entrambe le
condizioni patologiche?
I network cellulari correlano con la comorbidity
La Pearson correlation tra la comorbidity per tutte le coppie di malattie e
e n g, n p e ρ
φ
Correlazione positiva (PCC positivo) con tutte 3 variabili.
ng mostra la corelazione più alta.
LA CORRELAZIONE OTTENUTA NON E ALTISSIMA in quanto su
83.924 paia di malattie considerate per calcolare la Pearson correlation
solamente 2.239 paia sono legate da geni (658) oppure PPI (1.873)
Studio della comorbidity
La correlazione tra i network cellulari è positiva,
ma bassa
Limiti attuali dell’applicazione:
I valori ottenuti NON considerano:
- l ambiente
- lo stile di vita
- trattamenti vari
Non si conoscono tutte le associazioni gene-malattia
nel database OMIM
Esiste un certo noise tra l OMIM e i codici ICD-9-CM
I disease genes o le proteine condivise si conoscono
solamente per una minoranza delle malattie con comorbidity
I network cellulari correlano con la comorbidity
La comorbidity media per coppia di malattie che condividono
geni e proteine
<φ>
Se un paziente sviluppa una certa malattia associata con un gene o geni multipli hanno
2 volte maggiore chance di sviluppare un’altra malattia ad essa connessa nel diseasoma
http://pfam.sanger.ac.uk/
The Pfam database is one the most important collections of information in the world for classifying proteins. Its vision is to provide a tool which
allows experimental, computational and evolutionary biologists to classify protein sequences and answer questions about what they do and how
they have evolved. A 'periodic table' of biology
I network cellulari correlano con la comorbidity
La comorbidity media per coppie di malattie con valore crescente di ng, np, ρ
<φ>
Le malattie più interconnesse
hanno comorbidity più alta
<φ>
<φ>
3. Predizioni / Conclusioni
Studio della comorbidity
I network cellulari correlano con la comorbidity e predicono:
- Comorbidity note: diabete e obesità; cancro alla mammella e osteosarcoma
- Comorbidity nuove: malattia di Alzheimer e infarto di miocardio;
malattia del SN autonomo e la sindrome del tunnel carpale
Gene – proteina codificata
ACE-angiotensin-converting enzyme
APOE- apolipoproteina E
TTR-transthyretin
IKBKAP- IKAP protein
L impatto della topologia del network metabolico
sulla comorbidity delle malattie umane
Studio della comorbidity
Le vie metaboliche intracellulari correlano a
livello di popolazione con i pattern di comorbidity
delle malattie metaboliche?
Metabolic pathways
Metabolismo (dal greco µεταβολισµός (metabolismós) è il complesso delle
reazioni chimiche che avvengono in un organismo vivente o in una sua
parte con lo scopo di mantenere la vita.
E’ l'insieme dei processi di trasformazione chimica che avvengono
nell'organismo sia in rapporto all'assimilazione degli alimenti che alle
attività funzionali degli organi e dei tessuti.
"
"
Il metabolismo si divide in tre insiemi di processi:"
"
▪anabolismo, che produce molecole complesse "
a partire da molecole più semplici utili alla cellula;"
▪catabolismo, che comporta la degradazione di "
molecole complesse in molecole più semplici e produce "
energia; "
▪metabolismo energetico, che comporta il recupero "
dell'energia producendo molecole di ATP. "
La classe di malattie metaboliche è la più dispersa nella rete di malattie umane classificate in base ai geni condivisi
Perché fare la rete di malattie metaboliche?
1. Le mappe metaboliche conosciute sono di alta qualità e
sono abbastanza complete.
2. Vari sets di reazioni metaboliche consecutive sono
interdipendenti, e le loro attività (flux rate) a volte
sincronizzate. Quindi la disfunzione di una via si ripercuote
anche su altre vie metaboliche
3. Le cascate metaboliche suggeriscono che alcune
malattie metaboliche possono associarsi (comorbidity).
4. La classe di malattie metaboliche è la più dispersa nella
rete di malattie umane classificate in base ai geni condivisi
Duarte et al., 2007
Classificazione delle malattie metaboliche
Le malattie metaboliche ereditarie, sono classificabili nei seguenti gruppi:
Malattie del metabolismo degli acidi organici (metilmalonico, propionico,
isovalerico, glutarico acidemia, ecc.)
Malattie del metabolismo degli aminoacidi (fenilchetonuria, omocistinuria,
difetti del ciclo dell'urea, cistinuria)
Malattie del metabolismo dei carboidrati (glicogenosi, galattosemia, intolleranza
ereditaria al fruttosio, piruvato carbossilasi e deidrogenasi)
Malattie del metabolismo dei lipidi (ipercolesterolemia familiare e altre
dislipidemie)
Malattie del metabolismo delle purine e pirimidine (malattia di Lesch-Nyhan)
Malattie del metabolismo dei metalli (emocromatosi, malattia di Wilson, malattia
di Menkes)
Malattie mitocondriali (difetti della catena respiratoria, difetti dell'ossidazione
degli acidi grassi)
Malattie lisosomiali (mucopolisaccaridosi, malattia di Niemann-Pick, malattia di
Tay-Sachs, leucodistrofia metacromatica)
Malattie perossisomiali (adrenoleucodistrofia, malattia di Zellweger)
Malattie del metabolismo dei neurotrasmettitori (malattia di Canavan, difetto di
tirosina idrossilasi)
Le tappe dello studio della comorbidity delle
malattie metaboliche
1. Costruzione della rete di malattie metaboliche
2. Analisi e validazione della rete di malattie metaboliche
3. Analisi della comorbidity
4. Predizioni / Conclusioni
Le tappe dello studio della comorbidity delle
malattie metaboliche
1. Costruzione della rete di malattie metaboliche
a. KEGG, BiGG – lista generica, manually curated di reazioni
metaboliche
b. OMIM database per identificare le malattie associate con gli
ensimi del network metabolico umano (2,025 disease genes e
3,423 disease phenotypes - Aug. 2007)
c. Medicare database
Le tappe dello studio della comorbidity delle
malattie metaboliche
1. Costruzione della rete di malattie metaboliche
KEGG: 737 su 1.473 reazioni metaboliche sono associate a
malattie genetiche in OMIM
BiGG: 1.116 su 3.742 reazioni metaboliche sono associate a
malattie genetiche in OMIM
Table S1. Disease classification and disease-reaction association
Disorder class
Associated reactions
Ref[Goh
et al.,
2007]
cross
check
ICD-9-CM
& ICD-10CM
Final
(KEGG)
(BiGG)
17,20-lyase deficiency, isolated, 202110 (3)
Metaboli
c
Metaboli
c
Metaboli
c
R04853|R03783|R02211|R0
4852
P45017A3r|P45017A1r|P450
17A2r|P45017A4r
2-methyl-3-hydroxybutyryl-CoA dehydrogenase deficiency,
300438 (3)
Metaboli
c
Metaboli
c
Metaboli
c
R04743|R04745|R04748|R0
4737|R04739|R04741|R019
75|R04203|R05066|R06941
HACD9m|HACD1m
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
3-beta-hydroxysteroid dehydrogenase, type II, deficiency
(3)
Metaboli
c
Nonmetaboli
c
3-hydroxyacyl-CoA dehydrogenase deficiency, 231530 (3)
Metaboli
c
3-hydroxyisobutryl-CoA hydrolase deficiency, 250620 (3)
Metaboli
c
Disease
2-methylbutyrylglycinuria, 610006 (3)
3-Methylcrotonyl-CoA carboxylase 1 deficiency, 210200 (3)
3-methylglutaconic aciduria, type I, 250950 (3)
AGAT deficiency (3)
AICA-ribosiduria due to ATIC deficiency, 608688 (3)
Acatalasemia (3)
ACOAD10m|PPCOAOm|A
COAD1fm
R04138
MCCCrm
Metaboli
c
R02840|R04849|R03327|R0
3328|R02217|R04163|R041
64|R02842|R04851|R02499
|R02500|R01837|R01839|R
04678|R04680
HSD3B3r|HSD3B2r|HSD3B13r
|HSD3B12r|HSD3B11r
Metaboli
c
Metaboli
c
R04743|R04745|R04748|R0
4737|R04739|R04741|R019
75|R04203|R05066|R06941
HACD1m
Metaboli
c
Metaboli
c
3HPCOAHYD
R02085
ECOAH1m|T2M26DCOAHL
m|MGCHrm|PRPNCOAHY
Dm|C2M26DCOAHLm
R01989|R00565
GLYAMDTRc
R04560|R01127
AICART|IMPC
R02670|R00602
CAT2p|CATm|CATp
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
Metaboli
c
1
Principi per la costruzione della rete di malattie metaboliche:
Se lo stesso substrato è condiviso tra due reazioni metaboliche,
la sua scarsità o abbondanza influisce entrambe le vie
metaboliche
Due reazioni metaboliche sono linked se processano lo stesso
metabolita (per esempio sono adiacenti in una mappa metabolica)
Le reazioni metaboliche cellulari sono alla base dei links
(edges/archi) tra i nodi rappresentati da malattie metaboliche
1. Costruzione della rete di malattie metaboliche
http://
www.kegg.j
p/kegg/
kegg2.html
Pathway glicolisi
BPGM=bisfosfoglicerato mutasi
ENO= enolasi
PGAM=fosfoglicerato mutasi
Ipotesi di lavoro: se 2 malattie sono causate da diffetti metabolici che
influenzano “coupled reactions” (glycerate-2P), allora la loro patogenesi
potrebbe essere correlata (linked)
La rete delle malattie metaboliche umane (2008)
costruita usando KEGG database
- 308 nodi sono
connessi da
878 metabolic
links
Lee D et al. PNAS 2008;105:9880-9885
La rete delle malattie metaboliche umane (2008)
costruita usando BiGG database
319 malattie connesse da 699 metabilic links
La rete delle malattie metaboliche umane (2008)
costruita usando KEGG database
Lee D et al. PNAS 2008;105:9880-9885
2. Analisi e validazione della rete di malattie metaboliche (KEGG)
- Ha un cluster con 197 malattie (in network theory = “giant
component”) e altri clusters più piccoli
- Le malattie del giant component appartengono a classi
diverse
- Le malattie del metabolismo purinico formano un cluster
di 33 malattie (62 reazioni chimiche in KEGG) (blue shading)
- Le malattie del metabolismo lipidico formano un cluster di
34 malattie (34 reazioni chimiche in KEGG) (red shading)
- Hubs: ipertensione (27 links); resistenza alla warfarina (19) e
anemia emolitica (17).
- La maggior parte delle malattie hanno pochi links (la rete
contiene malattie mendeliane come la deficienza di enolasi ma anche malattie
complesse come ipertensione e diabete per le quali si conosce solamente
suscettibilità di alleli che non è suffiicente per indurre la malattia)
Distribuzione del degree e distanza nella rete di
malattie metaboliche
Degree medio = 5
Network bipartito: cerchi = malattie metaboliche; quadrattini = reazioni
metaboliche; links: disease-reaction; reaction-reaction
308 malattie
686 “reazioni”
1741 links malattia-reazione
431 links reazione-reazione
Validazione della rete: omogeneità funzionale
I links tra le malattie metaboliche e gli ensimi associati
hanno un significato funzionale? Gli ensimi sono espressi
negli stessi tessuti?
Per rispondere dobbiamo computare il PCC tra i profili di
espressione dei geni linked nei vari tessuti
Genomics. 2005 Aug;86(2):127-41.
Interpreting expression profiles of cancers by genome-wide survey of breadth of expression in
normal tissues.
Ge X, Yamamoto S, Tsutsumi S, Midorikawa Y, Ihara S, Wang SM, Aburatani H.
Genome Science Division, Research Center for Advanced Science and Technology, The University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo
153-8904, Japan. [email protected]
Abstract
A critical and difficult part of studying cancer with DNA microarrays is data interpretation. Besides the need for data
analysis algorithms, integration of additional information about genes might be useful. We performed genomewide expression profiling of 36 types of normal human tissues and identified 2503 tissue-specific genes.
We then systematically studied the expression of these genes in cancers by reanalyzing a large collection of
published DNA microarray datasets. We observed that the expression level of liver-specific genes in hepatocellular
carcinoma (HCC) correlates with the clinically defined degree of tumor differentiation. Through unsupervised
clustering of tissue-specific genes differentially expressed in tumors, we extracted expression patterns that are
characteristic of individual cell types, uncovering differences in cell lineage among tumor subtypes. We were able to
detect the expression signature of hepatocytes in HCC, neuron cells in medulloblastoma, glia cells in glioma, basal
and luminal epithelial cells in breast tumors, and various cell types in lung cancer samples. We also demonstrated
that tissue-specific expression signatures are useful in locating the origin of metastatic tumors. Our study shows
that integration of each gene's breadth of expression (BOE) in normal tissues is important for biological
interpretation of the expression profiles of cancers in terms of tumor differentiation, cell lineage, and metastasis.
Esempi di geni linked: ENO3 e PGAM; ENO3 e BPGM
Validazione della rete: omogeneità dinamica
La coespressione dei geni metabolici linked è maggiore
rispetto a geni non-linked
KEGG
P<10-8
PCC=0.66 e P=10-5 per ENO3 e PGAM2
BiGG
Considerazione: Le relazioni causali tra due malattie
metaboliche linked potrebbe non essere limitata
solamente a reazioni adiacenti
Ipotesi: Le malattie metaboliche potrebbero essere
linked da reazioni adiacenti, ma anche da reazioni “flux
coupled” (reazioni di flusso accoppiate)
Metabolic flux: the rate of passage of a metabolite (production/
consume) through a given metabolic pathway/reaction
Types of flux coupling between reactions that are located first
neighbour (directly connected by one node)
Set di
reazioni
correlate o
‘‘flux
coupled’’
i) A-B: directionally coupled, ii) B-C: fully coupled, and iii) C-D: uncoupled.
Le reazioni metaboliche di flusso riflettono associazioni
funzionali
Co-Regulation of Metabolic Genes Is Better
Explained by Flux Coupling Than by Network
Distance
Richard A. Notebaart1, Bas Teusink1,2,3, Roland J. Siezen1,2,3, Balázs Papp4,5*
1 Center for Molecular and Biomolecular Informatics (NCMLS), Radboud University Nijmegen Medical Center, Nijmegen, The Netherlands, 2 NIZO Food Research BV, Ede, The
Netherlands, 3 TI Food and Nutrition (WCFS), Wageningen, The Netherlands, 4 Institute of Biochemistry, Biological Research Center, Szeged, Hungary, 5 Faculty of Life
Sciences, The University of Manchester, Manchester, United Kingdom
To what extent can modes of gene regulation be explained by systems-level properties of metabolic networks? Prior
studies on co-regulation of metabolic genes have mainly focused on graph-theoretical features of metabolic networks
and demonstrated a decreasing level of co-expression with increasing network distance, a naı̈ve, but widely used,
topological index. Others have suggested that static graph representations can poorly capture dynamic functional
associations, e.g., in the form of dependence of metabolic fluxes across genes in the network. Here, we systematically
tested the relative importance of metabolic flux coupling and network position on gene co-regulation, using a
genome-scale metabolic model of Escherichia coli. After validating the computational method with empirical data on
flux correlations, we confirm that genes coupled by their enzymatic fluxes not only show similar expression patterns,
but also share transcriptional regulators and frequently reside in the same operon. In contrast, we demonstrate that
network distance per se has relatively minor influence on gene co-regulation. Moreover, the type of flux coupling can
explain refined properties of the regulatory network that are ignored by simple graph-theoretical indices. Our results
underline the importance of studying functional states of cellular networks to define physiologically relevant
associations between genes and should stimulate future developments of novel functional genomic tools.
La distribuzione del flusso nelle reti metaboliche riflette meglio le relazioni
funzionali Introduction
tra due geni implicati nel metabolismo
rispetto
alle
misure
[11,12], one might
expect that the
correlation
between
reaction fluxes across network states would provide a sound
topologicheIn della
rete
(distanza
tra
due
I geni
metabolici
“flux
recent years,
metabolic
networks of various
species
have nodi).
and biochemically
relevant
measure of functional
dependbeen reconstructed [1], and several systematic studies
ence between enzyme-encoding genes [13]. Therefore, we
coupled” (accoppiati
in regulation
reazioni
di flusso)
tendono ad essere coespressi e
addressed the issue of gene
in metabolism
[2–5].
hypothesized that dynamic functional associations (i.e.,
These studies have revealed important insights into trancorrelations) between fluxes, rather than static topological
legati funzionalmente.
scriptional regulation by integration of gene co-expression
properties of a metabolic network, could capture true
Citation: Notebaart RA, Teusink B, Siezen RJ, Papp B (2008) Co-regulation of metabolic genes is better explained by flux coupling than by network distance. PloS Comput Biol
4(1): e26. doi:10.1371/journal.pcbi.0040026
with historically defined modules (e.g., glycolysis) or with
graph-theoretical properties of reconstructed networks.
Although trends in gene co-regulation with network distance
functional associations between genes and consequently
would provide refined insights into the modes of transcrip-
cant biological insight as to which reactions are potentially missing from metabolic models, as well as which reactions may be
under coordinated regulation, alluding to a mechanism for the
continuous refinement of metabolic reconstructions through an
iterative model-building process. Specifically, stoichiometric
models of Escherichia coli metabolism utilized within the flux
balance analysis (FBA) framework have been used for (1) quali-
connectivity features of genome-scale metabolic networks. The framework is dem
modes for a given metabolic
environmental
condition and
hint
at Escherichia coli, and Saccharomyces cerevisiae
reconstructions
of organism
Helicobacter
pylori,
a more flexible metabolism (Stelling et al. 2002). This concept
determine whether any two metabolic fluxes, v1 and v2, are (1) directionally coupled, if a
has proven effective in the rational strain design for poly-!non-zero flux for v2 but not necessarily the reverse; (2) partially coupled, if a non-zero
hydroxybutyrate production in Saccaromyces
cerevisiae by quanthough
variable,
flux by
forthe
v2 addition
and viceofversa;
or (3) fully coupled, if a non-zero flux for v
tifying the additional
flexibility
gained
a nonbut alsoreaction
a fixed (Carlson
flux for et
v2 al.
and2002).
vice versa.
native transhydrogenase
Corre-Flux coupling analysis also enables the glo
reactions,
which
are allrefers
reactions
incapable
of carrying flux under a certain condition; e
spondingly, the set
of extreme
pathways
to the
minimum
the setofofdescribing
all possible
reactions whose
deletion forces the flux through a particular
set of flux vectorsascapable
all steady-state
flux distributions and areaffected
consequently
a subset
of elementary
modes
reactions
denoting
all reactions
whose fluxes are forced to zero if a particular
(Schilling et al. 2000).
As withthus
elementary
the number
of
approach
providesmodes,
a novel
and versatile
tool for aiding metabolic reconstru
extreme pathwaysmanipulations.
provides a measure of pathway redundancy.
The application of extreme pathway analysis has revealed that
Flux coupling finder (FCF) = Procedura per identificare set di
coupled e uncoupled reactions nelle reti metaboliche
3
These authors contributed equally to this work.
Corresponding author.
E-MAIL [email protected]; FAX (814) 865-7846.
Article and publication are at http://www.genome.org/cgi/doi/10.1101/
gr.1926504. Article published online before print in January 2004.
4
14:000–000 ©2004 by Cold Spring Harbor Laboratory Press ISSN 1088-9051/04; www.genome.org
Methods
[Supplemental material
is available
online at1www.genome.org.]
Genome
Research
www.genome.org
An overarching attribute of metabolic networks is their inherent
robustness and ability to cope with ever-changing environmental conditions. Despite this flexibility, network stoichiometry
and connectivity do establish limits/barriers to the coordination
and accessibility of reactions. The recent abundance of complete
genome sequences has enabled the generation of genome-scale
metabolic reconstructions for various microorganisms (Covert et
al. 2001; Price et al. 2003; Reed and Palsson 2003). These models
provide a largely complete
1,3
1,3
2 skeleton of the metabolic reactions
present in an organism. Examination of the structural and topological properties of metabolic networks is important at both the
1,4
conceptual level, to reveal the organizational principles of meta1
2
within
cellular
and at the practical
Department of Chemical Engineering, Pennsylvania State University, Universitybolic
Park,interactions
Pennsylvania
16802,
USA;networks,
Genomatica
level for more effectively focusing engineering interventions and
Inc., San Diego, California 92121, USA
ensuring the consistency of the underlying reconstructions.
To this end, the identification of blocked reactions (i.e., reactions
incapable ofthe
carrying
flux dueand
to the
stoichiometry of the
In this paper, we introduce the Flux Coupling Finder (FCF) framework
for elucidating
topological
flux
metabolic
network
under
steady-state
conditions)
and enzyme
connectivity features of genome-scale metabolic networks. The framework is demonstrated on genome-scale
subsets (i.e., groups of reactions that operate together in fixed
metabolic reconstructions of Helicobacter pylori, Escherichia coli, and Saccharomyces cerevisiae. The analysis allows one to
flux proportions under steady-state conditions) in metabolic
determine whether any two metabolic fluxes, v1 and v2, are (1) directionally
coupled,
a non-zeroconsiderable
flux for v1 implies
a in recent years
models
hasif attracted
interest
non-zero flux for v2 but not necessarily the reverse; (2) partially coupled,(Kholodenko
if a non-zeroetflux
for
v
implies
a
non-zero,
1
al. 1995; Rohwer et al. 1996; Pfeiffer et al. 1999;
though variable, flux for v2 and vice versa; or (3) fully coupled, if a non-zero
v1 implies
not only
a non-zero
Klamtflux
et al.for
2003).
The output
of these
analyses provides signifibut also a fixed flux for v2 and vice versa. Flux coupling analysis also enables
the global
identification
blocked
cant biological
insight
as to which of
reactions
are potentially missing condition;
from metabolic
models,
as well asdefined
which reactions may be
reactions, which are all reactions incapable of carrying flux under a certain
equivalent
knockouts,
under
coordinated
regulation,
alluding
to aofmechanism for the
as the set of all possible reactions whose deletion forces the flux through
a particular
reaction
to zero;
and sets
refinement
reconstructions
through an
affected reactions denoting all reactions whose fluxes are forced to zerocontinuous
if a particular
reactionofismetabolic
deleted. The
FCF
iterative
model-building
stoichiometric
approach thus provides a novel and versatile tool for aiding metabolic
reconstructions
andprocess.
guidingSpecifically,
genetic
models of Escherichia coli metabolism utilized within the flux
manipulations.
balance analysis (FBA) framework have been used for (1) quali-
Flux Coupling Analysis of Genome-Scale Metabolic
Network Reconstructions
Anthony P. Burgard,
Costas D. Maranas
Evgeni V. Nikolaev,
Christophe H. Schilling, and
An overarching attribute of metabolic networks is their inherent
robustness and ability to cope with ever-changing environmental conditions. Despite this flexibility, network stoichiometry
and connectivity do establish limits/barriers to the coordination
and accessibility of reactions. The recent abundance of complete
genome sequences has enabled the generation of genome-scale
metabolic reconstructions for various microorganisms (Covert et
al. 2001; Price et al. 2003; Reed and Palsson 2003). These models
provide a largely complete skeleton of the metabolic reactions
present in an organism. Examination of the structural and topo-
3
These authors contributed equally to this work.
Corresponding author.
E-MAIL [email protected]; FAX (814) 865-7846.
Article and publication are at http://www.genome.org/cgi/doi/10.1101/
gr.1926504.
Article published
online
before print
in January 2004.
tatively predicting
the outcomes
of gene
knockout
experiments
[Supplemental material is available online at www.genome.org.]
4
tatively predicting the outc
(Edwards and Palsson 2000;
tifying the correct sequenc
creasingly anaerobic condit
tatively predicting cellular
Ibarra et al. 2002); (4) assess
bolic networks in response
gard and Maranas 2001); an
egies for enhancing biochem
Pharkya et al. 2004).
In the postgenomic er
entity, or physiological even
network of interactions. Foll
for examining structural a
based on convex analysis
strated for small-scale metab
to identify extreme pathway
modes (Schuster and Hilge
elementary mode refers to a
operate under steady-state c
tion can be represented by a
elementary modes. Element
quantitative measure of netw
modes for a given environm
a more flexible metabolism
has proven effective in th
hydroxybutyrate production
tifying the additional flexibi
native transhydrogenase re
spondingly, the set of extre
set of flux vectors capable o
tributions and are consequ
(Schilling et al. 2000). As wi
extreme pathways provides
The application of extreme
(Edwards and Palsson 2000; Badarinarayana et al. 2001); (2) identifying the correct
sequence
of Cold
byproduct
secretion
underPress
in- ISSN 1088-9051/04; www.genome.org
14:000–000
©2004 by
Spring Harbor
Laboratory
creasingly anaerobic conditions (Varma et al. 1993); (3) quantitatively predicting cellular growth rates (Edwards et al. 2001;
Ibarra et al. 2002); (4) assessing the performance limits of metabolic networks in response to gene additions or deletions (Burgard and Maranas 2001); and (5) suggesting gene knockout strategies for enhancing biochemical production (Burgard et al. 2003;
Pharkya et al. 2004).
La coespressione dei geni metabolici flux-coupled
è maggiore rispetto ai geni linked in reazioni adiacenti
Average PCC:
geni flux-coupled = 0.31
geni di reaz. adiacenti = 0.24
tutti i geni = 0.1
2. Analisi e validazione della rete di malattie metaboliche
Conclusioni:
1. Le malattie metaboliche linked da reazioni chimiche
formano una rete scale-free
2. Links funzionali tra i geni adiacenti (linked) e flux-coupled
suggeriscono l’esistenza di links tra malattie e comorbidity
Sarà vero?
3. Analisi della comorbidity
Scopo principale: capire se i links della rete di malattie
metaboliche possono essere detettati/individuati
realmente nella popolazione umana, quindi se le malattie
linked coesistono anche nei pazienti con malattie
metaboliche (o “metabolism-related” – correlate al
metabolismo)
3. Analisi della comorbidity
"The International Classification of Diseases, 9th Revision, Clinical
Modification" (ICD-9-CM), issued for use beginning October 1, 2008 for federal
fiscal year 2009 (FY09). http://www.cms.gov/
http://www.cms.gov/MedicareGenInfo/
http://www.medicare.gov/
The ICD-9-CM is maintained jointly by the National Center for Health Statistics
(NCHS) and the Centers for Medicare & Medicaid Services (CMS).
http://www.icd9data.com/
001-139
140-239
240-279
280-289
290-319
320-389
390-459
460-519
520-579
580-629
630-679
680-709
710-739
740-759
760-779
780-799
800-999
INFECTIOUS AND PARASITIC DISEASES
NEOPLASMS
ENDOCRINE, NUTRITIONAL AND METABOLIC DISEASES, AND IMMUNITY DISORDERS
DISEASES OF THE BLOOD AND BLOOD-FORMING ORGANS
MENTAL DISORDERS
DISEASES OF THE NERVOUS SYSTEM AND SENSE ORGANS
DISEASES OF THE CIRCULATORY SYSTEM
DISEASES OF THE RESPIRATORY SYSTEM
DISEASES OF THE DIGESTIVE SYSTEM
DISEASES OF THE GENITOURINARY SYSTEM
COMPLICATIONS OF PREGNANCY, CHILDBIRTH, AND THE PUERPERIUM
DISEASES OF THE SKIN AND SUBCUTANEOUS TISSUE
DISEASES OF THE MUSCULOSKELETAL SYSTEM AND CONNECTIVE TISSUE
CONGENTIAL ANOMALIES
CERTAIN CONDITIONS ORIGINATING IN THE PERINATAL PERIOD
SYMPTOMS, SIGNS, AND ILL-DEFINED CONDITIONS
INJURY AND POISONINGRY AND POISONING
http://www.icd9data.com/2011/Volume1/140-239/170-175/170/default.htm
2011 ICD-9-CM Diagnosis Code 170
Malignant neoplasm of bone and articular cartilage
2011 ICD-9-CM Diagnosis Code 170.0
Malignant neoplasm of bones of skull and face except mandible
2011 ICD-9-CM Diagnosis Code 170.1
Malignant neoplasm of mandible
2011 ICD-9-CM Diagnosis Code 170.2
Malignant neoplasm of vertebral column excluding sacrum and coccyx
2011 ICD-9-CM Diagnosis Code 170.3
Malignant neoplasm of ribs sternum and clavicle
2011 ICD-9-CM Diagnosis Code 170.4
Malignant neoplasm of scapula and long bones of upper limb
2011 ICD-9-CM Diagnosis Code 170.5
Malignant neoplasm of short bones of upper limb
2011 ICD-9-CM Diagnosis Code 170.6
Malignant neoplasm of pelvic bones sacrum and coccyx
2011 ICD-9-CM Diagnosis Code 170.7
Malignant neoplasm of long bones of lower limb
2011 ICD-9-CM Diagnosis Code 170.8
Malignant neoplasm of short bones of lower limb
2011 ICD-9-CM Diagnosis Code 170.9
Malignant neoplasm of bone and articular cartilage site unspecified
Studio della comorbidity
Studio della comorbidity
Studio effettuato su pazienti registrati nel
database Medicare:
N= 13.039.018 pazienti registrati su Medicare per un totale di
32.341.348 visite mediche
Periodo: più di 4 anni (1990 to 1993)
Età media: 76.3 ± 7.4 (range 65-113)
41.8% maschi; 89.9% caucasici
Caratteristiche del data-set Medicare
Nr. medio di
malattie/pz = 8,4
Costruzione di una mappa “hand-curated” per omologare le malattie
genetiche (OMIM) ai codici ICD-9-CM del database Medicare
Table S4. Disease prevalence
Disease
Prevalenza di una
malattia i:
Ii=Ni/N
(la frazione di
pazienti che
hanno quella
malattia)
Ni= nr. pz.
diagnosticati con
la malattia i
N= nr. totale di
pazienti
17,20-lyase deficiency, isolated, 202110 (3)
2-methyl-3-hydroxybutyryl-CoA dehydrogenase deficiency,
300438 (3)
2-methylbutyrylglycinuria, 610006 (3)
3-Methylcrotonyl-CoA carboxylase 1 deficiency, 210200 (3)
3-beta-hydroxysteroid dehydrogenase, type II, deficiency (3)
3-hydroxyacyl-CoA dehydrogenase deficiency, 231530 (3)
3-hydroxyisobutryl-CoA hydrolase deficiency, 250620 (3)
3-methylglutaconic aciduria, type I, 250950 (3)
AGAT deficiency (3)
AICA-ribosiduria due to ATIC deficiency, 608688 (3)
AMP deaminase deficiency, erythrocytic (3)
Abruptio placentae, susceptibility to (3)
Acatalasemia (3)
Achondrogenesis Ib, 600972 (3)
Acquired long QT syndrome, reduced susceptibility to,
152427 (3)
Acromesomelic dysplasia, Hunter-Thompson type, 201250 (3)
Acyl-CoA dehydrogenase, long chain, deficiency of, 201460
(3)
Adenosine deaminase deficiency, partial, 102700 (3)
Adenylosuccinase deficiency, 103050 (3)
Adrenal hyperplasia, congenital, due to 11-beta-hydroxylase
deficiency, 202010 (3)
Adrenal insufficiency, congenital with or without 46, XY sex
reversal (3)
Adult i phenotype with congenital cataract, 110800 (3)
Agammaglobulinemia, 601495 (3)
Agenesis of the corpus callosum with peripheral neuropathy,
218000 (3)
Albinism, brown oculocutaneous, (3)
ICD-9CM
Prevalence
code
282.3 4.44819E-06
277.85 0.00000E+00
277.85 0.00000E+00
277.85 0.00000E+00
255.2 1.46483E-05
277.85 0.00000E+00
0.00000E+00
277.86 0.00000E+00
282.3 4.44819E-06
277.6 4.33315E-05
279.2 1.09671E-05
0.00000E+00
277.89 0.00000E+00
733 4.58035E-02
426.82 0.00000E+00
759.2
7.67696E-05
277.85 0.00000E+00
277.2
277.2
1.76394E-06
1.76394E-06
255.2
1.46483E-05
0.00000E+00
270.3
279
2.30079E-06
6.21519E-04
742.2
1.07370E-05
270.2
1.48784E-05
193 coppie di malattie con comorbidity statisticamente
significativa
!"#$%&'()&*+,%",%&-"+.,&/0"/&0"1%&"&,+23+4+5"3/&5676.#+8+/9&"38&".%&5633%5/%8&+3&%+/0%.&:;<<&6.&=+<<&8"/"#",%&
23$*4$5&
*4&"
.67&48&5"
43$*4$5&*4&"
#$%&'%&!"
2@A#B":'8&"3C"5&4?$*&"3C"?>*D"C>*48$3*"$*B"
EFEGEH""
23:3*':<"':8&:<"5$%&'%&"$*"C'9$?$'?"
I<7&:4I3?&%8&:3?&9$'B"7:38&48$3*"'D'$*%8B"
!NHOGF""
#$%&'%&("
)$*+,-.//0"
)$*+,1$//0"
.97I<%&9'""
23**&48&5"
#$%43**&48&5"
!JK!(G"
K(KFGLGGHH"
!LE(JFH.MF!"
FLNFHGFHGHE"
P<7&:8&*%$3*B"5$'%83?$4B":&%$%8'*4&"83B"
EFOE((""
23**&48&5"
#$%43**&48&5"
!N(KGN!"
!FO!OFELHKG"
!L(GKGG.MF!"
FLOF(K!((NN"
P&93?<8$4"'*&9$'""
Q7I&:34<83%$%B"I&:&5$8':<""
@;&%$8<B"'5:&*'?"$*%>CC$4$&*4<B"'*5"
:&5"I'$:""
#$%43**&48&5"
23**&48&5"
!FHK"
!FL(GHHJOKO"
OLOKHJ!.MF("
FLE(JEOJGO!"
23**&48&5"
23**&48&5"
!!KEHO"
KH!KFLEJGEJ"
OLH(EKE.MF("
FLHKH!KJ(FE"
#$';&8&%"9&??$8>%B"D&%8'8$3*'?B"!(KOK!""
P<7&:8&*%$3*B"5$'%83?$4B":&%$%8'*4&"83B"
EFOE((""
2393:;$5$8<"
='6$9>9"
73%%$;?&"
4393:;$5$8<"
23:3*':<"%7'%9%B"%>%4&78$;$?$8<"83""
23**&48&5"
23**&48&5"
H(EK!H"
((KEHJLGNK("
JLN!NFO.MF("
FLH(EEJ(FKG"
23*&"5<%8:37I<M!B"HFNF(F""
)&;&:"43*D&*$8'?"'9'>:3%$%"RB"(FNFFF""
23**&48&5"
23**&48&5"
!"
(L(!.MFK"
KLOG(NH.MF("
FLGN(OFOGEG"
)&;&:"43*D&*$8'?"'9'>:3%$%"RB"(FNFFF""
S&8$*'?"43*&"5<%8:37I<"HB"E!FF(N""
23**&48&5"
23**&48&5"
!"
(L(!.MFK"
KLOG(NH.MF("
FLGN(OFOGEG"
P&93%$5&:3%$%B"%<%8&9$4B"5>&"83"
'4&:>?37?'%9$*&9$'B"EFN(GF""
23**&48&5"
#$%43**&48&5"
HHNO("
!(N!HLJK(OG"
KLKF(O(.MF("
FLHKHN(HJKN"
23**&48&5"
23**&48&5"
GNKK"
!ONOLEOK!(H"
KLFHJ((.MF("
FL((KE(ONFG"
#$%43**&48&5"
23**&48&5"
((KN"
!NELOJ!JN!K"
NLOOJHH.MF("
FL!!NFJ(NGO"
PUSA"%<*5:39&B"EFJ(HE""
R:3*"5&C$4$&*4<"'*&9$'B"%>%4&78$;$?$8<"
83""
T38'?"$35$5&"3:D'*$C$4'8$3*"5&C&48B"
(JNKFF""
@;%&%%$V&M4397>?%$V&"5$%3:5&:"!B"
!EN(HF""
@;&%$8<B"'5:&*'?"$*%>CC$4$&*4<B"'*5"
:&5"I'$:""
23**&48&5"
#$%43**&48&5"
!EJKO"
KOEJLKFGOOK"
NLF(HNO.MF("
FLOEEEJKENN"
U?53%8&:3*$%9B"D?>4343:8$43$5M
:&9&5$';?&B"!FHGFF""
U77':&*8"9$*&:'?343:8$43$5"&64&%%B"
I<7&:8&*%$3*"5>&"83""
23**&48&5"
#$%43**&48&5"
EK"
FL(NONOFOF!"
HLKGJO(.MF("
FLG!KKNOG!!"
#$';&8&%"9&??$8>%B"D&%8'8$3*'?B"!(KOK!""
PUSA"%<*5:39&B"EFJ(HE""
P<7&:8&*%$3*B"5$'%83?$4B":&%$%8'*4&"83B"
EFOE((""
23**&48&5"
#$%43**&48&5"
EH!GE"
NF!EELO(KKH"
HLK!G(H.MF("
FLHFEFJ(JNG"
23**&48&5"
#$%43**&48&5"
G!KOJ"
EEKOKL!!!!G"
HLH!GOG.MF("
FL!JN!EH!GH"
W':C':$*":&%$%8'*4&B"!((JFF""
23**&48&5"
#$%43**&48&5"
HOE!K"
(HKJKLFGGJN"
HL!HE(G.MF("
FL!KFHEKFH"
@V':$'*"4'*4&:""
2I3?&%8'%$%B";&*$D*":&4>::&*8"
$*8:'I&7'8$4B"(NHHFF""
#$';&8&%"9&??$8>%B"D&%8'8$3*'?B"
!(KOK!""
Q4I$X3'CC&48$V&"5$%3:5&:B"
%>%4&78$;$?$8<"83B"!O!KFF""
23**&48&5"
23**&48&5"
!HKG"
!(OLO(NONK!"
HLF!!NH.MF("
FLJHOEH!!KH"
#$%43**&48&5"
23**&48&5"
(KOG"
NHNLJ(GKNKN"
(LOOJG!.MF("
FLHK!HFOHNG"
23**&48&5"
#$%43**&48&5"
(!JFJJ"
!OHFHHLO((K"
(LK((JK.MF("
FLEJEHKKFH("
23**&48&5"
#$%43**&48&5"
(NOG"
K!GLHHOEH(H"
(LN!JK!.MF("
FLHKJGJ(OEE"
P<7&:8I<:3$5$%9B"43*D&*$8'?""
U*6$&8<M:&?'8&5"7&:%3*'?$8<"8:'$8%B"EFJOHN""
Coincidence/Co-occurence = Nr. pz. che presentano le due malattie (Cij)
Expected coincidence = coincidenza attesa se le due malattie
fossero indipendenti (Cij* = IiIj/N)
Comorbidity index (vedi φ correlation)
Maximum possible comorbidity = comorbidity se la coincidenza fosse
uguale all’incidenza più bassa delle due malattie
/?'>439'"!B"37&*"'*D?&B".B"!HJJEF""
23:3*':<"':8&:<"5$%&'%&"$*"C'9$?$'?"
I<7&:4I3?&%8&:3?&9$'B"7:38&48$3*"'D'$*%8B"
!NHOGF""
.*539&8:$'?"4':4$*39'""
/'??;?'55&:"5$%&'%&"!B"EFFOFH""
U*&9$'B"#$'93*5M1?'4+C'*B"!FKEKF""
A':+$*%3*"5$%&'%&"!HB"E!F(GJ""
Le malattie metaboliche con la più alta comorbidity
Table S6. Disease pairs that have the highest comorbidity and are connected in both KEGG and BiGG
database
Coincidence
Disease1
Disease2
Diabetes mellitus
Hypertension
Hyperthyroidism, congenital
Endometrial carcinoma
Glutathione synthetase deficiency
Lhermitte-Duclos syndrome
Alcoholism, susceptibility to
Goiter
Goiter
Diabetes mellitus
Enolase-beta deficiency
Aldosteronism, glucocorticoid-remediable
Favism
Asthma
Obesity
Coronary spasms, susceptibility to
Total iodide organification defect
Ovarian cancer
Myocardial infarcation, susceptibility to
Oligodendroglioma
Epilepsy
Hyperthyroidism, congenital
Total iodide organification defect
Hyperinsulinemic
hypoglycemia,
familial, 3
Myopathy
Hypoaldosteronism, congenital
Hemolytic anemia
Atopy
Aldosteronism, glucocorticoid-remediable
Asthma
Glutathione synthetase deficiency
Low renin hypertension, susceptibility to
Atherosclerosis, susceptibility to
Hemolytic anemia
1148 (662)
7084 (5889)
210 (80)
Colon adenocarcinoma
Diabetes mellitus
Ovarian cancer
Hemolytic anemia
816 (505)
1656 (1215)
Colon adenocarcinoma
Cowden disease
93 (25)
(expected value)
115638 (53151)
326513 (225637)
9455 (1849)
1359 (129)
4900 (1725)
109 (3)
2038 (656)
426 (52)
2489 (977)
711 (175)
107 (7)
58 (3)
13 (0.2)
341 (90)
Comorbidity
(maximum possible
comorbidity)
8.3266!10-2 (0.35316)
7.4141!10-2 (0.32667)
5.0372!10-2 (0.22563)
3.0114!10-2 (0.73863)
2.1414!10-2 (0.95010)
1.6810!10-2 (0.80107)
1.5058!10-2 (0.58752)
1.4343!10-2 (0.72645)
1.3767!10-2 (0.16397)
1.2320!10-2
(0.019991)
1.0519!10-2 (0.37852)
9.0245!10-3 (0.32134)
7.8440!10-3 (0.22424)
7.4047!10-3 (0.11757)
6.4234!10-3
(0.017235)
4.4056!10-3 (0.96029)
4.0737!10-3 (0.22471)
3.8768!10-3
(0.370902)
3.8373!10-3 (0.05276)
3.8059!10-3
(0.082376)
1
Le malattie metaboliche connesse hanno alta comorbidity
Average comorbidity:
- Malattie metaboliche: 0.0009
- Malattie metaboliche connesse: 0.0027
- Malattie metaboliche “flux connected”: 0.0062
La prevalenza (Ix) delle malattie metaboliche
Ix = prevalenza di una malattia x (la frazione di pazienti che
hanno quella malattia)
Comorbidità e prevalenza nella rete di malattie metaboliche
Le prevalenze più alte:
Ipertensione: 0.337
Malattia coronarica: 0.246
Diabete mellito: 0.167
Malattia polmonare: 0.147
Lee D et al. PNAS 2008;105:9880-9885
La prevalenza delle malattie metaboliche aumenta con
il degree nella rete
Più una malattia è connessa nella rete più aumenta la probabilità
che venga indotta da altre malattie e che contribuisca
all’insorgenza di altre malattie (comorbidity)
La comorbidity di due malattie metaboliche decresce con
l’aumento della loro distanza nella rete
Distanza: lunghezza (numero di links) della shortest path tra
due nodi
Le malattie più connesse hanno una rata della mortalità più alta
Un paziente con una malattia hub ha la tendenza di sviluppare
le malattie conesse nella rete (che aumentano la mortalità)
I links tra le reazioni chimiche predicono la
comorbidity delle malattie metaboliche meglio rispetti
ai geni condivisi
PCCs tra la presenza
di geni condivisi e
links metabolici con
disease comorbidity
sono presentati per le
malattie correlate al
metabolismo e le
malattie metaboliche
classiche.
Geni condivisi
Links metabolici (KEGG)
Links metabolici (BiGG)
Malattie
con link
metabolico
e senza link
genico
I network cellulari metabolici correlano con la comorbidity
Tuttavia solamente 31% dei nodi linked della rete
mostrano la tendenza di co-esistere nella popolazione.
Limiti:
Solamente 46% delle malattie entrano nella categoria
“metaboliche”
I valori ottenuti NON considerano:
- l ambiente
- lo stile di vita
- trattamenti vari dei pazienti
- che le reti metaboliche potrebbero avere specificità cellulari
La rete di malattie metaboliche si basa sulle informazioni trovate
sull’OMIM. Tuttavia NON si conoscono tutte le associazioni genemalattia nel database OMIM
Conclusioni (1):
1. Le malattie metaboliche si organizzano in una rete
se gli ensimi coinvolti sono linked in vie metaboliche
2. I migliori “predictors” di link tra malattie sono le
reazioni chimiche delle vie metaboliche
3. Le malattie metaboliche si organizzano in una rete
complessa scale free
Conclusioni (2):
1. Le malattie linked/connesse nella rete presentano
maggiore comorbidity rispetto a quelle non connesse
2. Il degree alto delle malattie (nodi) predice una alta
prevalenza nella popolazione
3. L’occorrenza di una malattia in un paziente aumenta la
probabilità di sviluppare le malattie linked nella rete.
Conclusioni (3)
1. L’approccio basato sull’utilizzo della rete è utile in quanto
predice la comorbidity delle malattie metaboliche o correlate
2. Le vie metaboliche cellulari si propagano a livello di
popolazione e predicono la comorbidity delle malattie
3. L’informazione codificata nella struttura delle reti
metaboliche si amplifica e diventa individuabile a livello della
popolazione come patterns di comorbidity
