Systems Biology (Methods in Molecular Biology)

been developed to infer biochemical networks from high-

throughput data [18, 19]. Despite their importance, these techni-

ques face many challenges, for example false positive and false

negative interactions are one of the main challenges that influence

the results of inference. On the other hand, machine learning

algorithms for inference are complex; but simple methods, like

Naive Bayes, may not work well for complex situations, they are

slow to train and prone to overfit [19]. A more detailed and highly

focused network, centered around particular disease [20, 21]or

process [22–24], can be constructed by expert domain knowledge

(functional and structural information), diligent manual search for

published literature, and publically available databases (seeTable 1

for some of available databases for retrieving interaction for differ-

ent types of biochemical networks). To avoid the laborious manual

curation for network construction, some methods are developed to

automatically reconstruct networks by retrieving interactions or

sub-networks from existing maps and models [25, 26]. Combining

automatic reconstruction with domain knowledge, manual search

of literature and databases would provide a reasonable strategy to

construct detailed and fully annotated large-scale biochemical net-

works (seeFig. 3).

All these maps are a formalized representation of information

that can subsequently be analyzed with computational algorithms.

They are organized interactions knowledge-bases which help in:

(1) Gathering disperse information about complex biological sys-

tems at one place. (2) Managing and organizing information in a

standard pathway diagram format that is helpful to conceptually

analyzeand intuitivelyvisualizethenetworkcomponents. (3)Provide

information about the interactions to develop hypothesis that can

experimentally be tested. (4) Provide a foundation to derive simula-

tion models to analyze the dynamics of interacting components.

Their structural analysis allows the identification of functional mod-

ules [1–3], regulatory motifs [4, 5], and hub nodes [5, 27–29] along

Table 1
Important databases for retrieving interactions for various biochemical interactions

Type of network Databases

Gene regulatory networks

(GRN)

KEGG, TRANSFAC, TRED, TransPath

Metabolic networks KEGG, BioCyc, MetaCyc, BRENDA, BiGG, metaTIGER

Protein-protein

interaction (PPI)

BioGrid, HPRD, STRING, IntAct, DIP, MIPS

Signal transduction

networks (STN)

PID, BioCarta, SPIKE, WikiPathways, CST Signaling Pathways, The Cell

Collective, iHOP, SignaLink, NetPath

MicroRNA interaction

network

miRecords, TarBase, miRTarBase, miRWalk, miRGen, TransmiR, UCSC

browse

252 Faiz M. Khan et al.