Genomics in Drug Discovery
Download
Report
Transcript Genomics in Drug Discovery
Genomics in Drug Discovery
@ Organon, Oss
2005-08-22
Tim Hulsen
Introduction
• Proteins are vital to life:
involved in all kinds of life
processes
• Understanding protein functions
and relationships is very important for
drug design
• Currently, the molecular function of about
40% of the proteins is unknown
Introduction
Availability of fully sequenced genomes gives us a wealth
of information:
currently more than 15 eukaryotic genomes have nearly
been completely sequenced, over 148 microbial genomes
and over 1000 viruses.
Determine protein function by using different in silico
techniques:
• sequence comparison to known protein sequences
• sequence clustering with proteins which have the same
or similar function
Genomics @ Organon:
The Protein World project
• All-against-all sequence comparison of
complete proteomes from 145 species
• Smith-Waterman algorithm + Z-value
(Monte-Carlo statistics)
Protein World and its ambitions
Build and maintain a sequence similarity repository of all
complete proteomes and aligning it with “omics”
research in the Netherlands
Classification of all proteins into groups of related proteins
• A phylogenetic repository
• Annotation of new sequences
• Mining protein families
• Identification of genes common / specific to (groups of)
species
Applications of Protein World
Structural properties
• Protein comparison coupled to structure related databases
(PDB, SCOP, etc.)
Systems biology
• Connecting PW to other databases (pathways, protein
interactions, literature etc.)
Orthology
• Annotation of new proteins
• To predict discrepancies and similarities between species
Orthology
• Describes “the evolutionary relationship
between homologous genes whose
independent evolution reflects a speciation
event” (Fitch, 1970)
Protein World & Drug Discovery
• Orthologies can be used to transfer
function of proteins in model organisms
(mice, rats, dogs, etc.) to humans
• Drugs tested on model organisms can
have different effects in humans. Why?
• Could be explained by looking at proteins
in drug pathways and their orthologs
• Example: trypsin inhibition pathway
Trypsin inhibition pathway (1)
• Organon: thrombin inhibitors
• Needed to stop thrombosis (blood clotting)
• Thrombin inhibitor on the market: (xi)melagatran, sold as
Exanta (AstraZeneca)
• Proven to be better than warfarin, but …
Trypsin inhibition pathway (2)
• Side effect of thrombin inhibitors: inhibition
of trypsin
• Trypsin inhibition -> rise in cholecystokinin
(CCK) levels -> stimulation of pancreas ->
pancreatic tumors
• Difficult to test in model organisms:
– Rat: very strong CCK response
– Mouse: weak CCK response
– Human: almost no CCK response
Trypsin inhibition pathway (3)
Trypsin inhibition pathway (4)
Ortholog identification methods:
1. Using functional annotation (SPTrEMBL):
2. Best Bidirectional Hit (BBH)
one-to-one relationships
3. PhyloGenetic Trees (PGT)
many-to-many relationships
Best Bidirectional Hit (BBH)
• Very easy and quick
• Human protein (1) SW best
hit in mouse/rat (2)
• Mouse/rat protein (2) SW
best hit in human (3)
• If 3 equals 1, the human and
mouse/rat protein are considered
to be orthologs
PhyloGenetic Tree (PGT)
PROTEOME
Human
PROTEOMES All
SELECTION OF HOMOLOGS
eukaryotic
LIST Hs-Mm pairs
Hs-Rn pairs
proteomes
ALIGNMENTS AND TREES
TREE SCANNING
PHYLOME
Z>20
RH>0.5*QL
~25,000 groups
Trypsin inhibition pathway (5)
Mm – Hs – Rn
- by annotation
- BBH
- PGT
Trypsin inhibition pathway (6)
• PGT method: in some cases too many orthologous
relationships, especially for trypsin (73 in mouse and 62 in
rat!)
• BBH method seems to be more usable for this study, but still
not gives an explanation for the differences in CCK levels
• Our problem (different CCK responses in Human, Mouse
and Rat) cannot be solved only by orthology identification
• Combine ortholog analysis with other data
• Focus on the molecules that are most likely to be
responsible for these differences: CCK and trypsin
Trypsin inhibition pathway (7)
• Current activities:
– Take a better look at regulation:
promoter detection?
– Use expression data?
– Structural explanation? Modelling of
interactions between the involved
molecules
Possible student projects
• Orthology case study: explain differences
between humans and model organisms
(like example of trypsin inhibition pathway)
• Chicken project (in collaboration with
Wageningen University): comparison of immune
system in chickens to i.s. in humans and other
vertebrates
• Cluster algorithms
People
•
•
•
•
Peter Groenen
Wilco Fleuren
Tim Hulsen
Others @ MDI
• Students?