Computational drug target gene discovery

CREATED: 200809290400 Speaker: Yoshinori Tamada ** Drug target discovery Input is knockout microarray and drug specific microarray

Drug affected genes

** model drugs as virtual gene (virtual gene technique) ** use boolean network

Druggable genes

** genes which regulate drug affected gene ** use Bayesian network ** Drug active pathway discovery Strategy: Knockout micoarray $\rightarrow$ Gene network (Bayesian network) $\rightarrow$ Discrete Bayesian Network $\rightarrow$ Identify drug active pathway

• Drug active score (a node is either drug active or parent active)
• Time expanded network ** Drug target discovery on human cells
• Druggable gene networks: a gene regulatory network affected by a drug, contains known drug target genes and their regulatory pathways
• Time-course data $\rightarrow$ dynamic Bayesian model to estimate dynamic relationships
• Knock-down data $\rightarrow$ possible regulatory relations
• Bayesian data fusion for combining three types of information to estimate druggable gene network