Transcript Merck`s Estimated loses from Law suits regarding Vioxx

Predicting Toxicity
-An Unfinished Case Story
Ph.d. Student Jeppe Skytte Spicker
Merck’s Estimated loses from Law suits
regarding Vioxx:
50 BILLION DOLLARS!!
(Jyllands-Posten 14. december 2005, erhverv og Økonomi, side 6)
Drug profiling
Database on “profiles” of
known compounds
Drug profiling
New, unknown
compound
Project aims:
1. Make a prediction method or a
classifier for toxicity
2. Gain mechanistic insight of
toxicity
Assignment (work):
•
You have tissue from rats treated with 15 different
compounds.
–
–
–
•
You have pathology reports on toxicological findings in the rat.
Five of the compounds do not induce a toxic response, the
remaining ten does.
The ten remaining compounds all induces various forms of
toxicity which all manifests as cell death in the liver.
You have 10 minutes to formulate your approach and
some interesting questions. Afterwards we will discuss
your suggestions in forum.
Questions:
•
•
•
•
•
•
Is it possibly to predict? (good performance)
Can I separate subclasses?
How few genes are needed?
(How large is the fold change?)
Do these genes make biological sense?
Can I get any mechanistic insight by putting the
genes into biological context (pathways)?
Approach to classifier:
• Make independent feature selection (ttest/ANOVA)
• KNN LGO (leave group out)
FINDING THE SIGNIFICANT GENES
IS ONLY 1% OF THE WORK!
•
•
•
•
Find an expert on the biology (It is not a
weakness if that expert comes from a completely
different planet)
Make sure that person is (as) interested in the
problem (as you) (Identify a win-win situation)
Ask the person if what you are looking at is
interesting and relevant FROM THE VERY
BEGINNING!
Never stop looking for future partners (for
collaboration)
Which list do you like?
Probe Set ID
Gene Title
GO Biological Process Description
1368399_a_at
plasma glutamate carboxypeptidase
proteolysis and peptidolysis
peptide metabolism
tissue regeneration
1368609_at
solute carrier family 10
(sodium/bile acid cotransporter
family), member 1
transport
sodium ion transport
organic anion transport
bile acid transport
1371400_at
thyroid hormone responsive protein
regulation of transcription from RNA polymerase II promoter
lipid metabolism
1372306_at
ethylmalonic encephalopathy 1
(predicted)
1387053_at
flavin containing monooxygenase 1
electron transport
transport
1387328_at
Cytochrome P450, subfamily IIC
(mephenytoin 4-hydroxylase)
electron transport
xenobiotic metabolism
Or?
Probe Set ID
Gene Title
GO Biological Process Description
1369983_a_at
plasma glutamate carboxypeptidase
proteolysis and peptidolysis
peptide metabolism
tissue regeneration
1003714_at
solute carrier family 6
(sodium/bile acid cotransporter
family), member 6b
transport
sodium ion transport
organic anion transport
bile acid transport
1603689_at
HSP90 (Heat shock protein)
response to cellular stress
1337206_at
CYP2B6
xenobiotic metabolism
lipid metabolism
1053873_at
flavin containing monooxygenase 1
electron transport
transport
1387328_at
Cytochrome P450, subfamily IIC
(mephenytoin 4-hydroxylase)
electron transport
xenobiotic metabolism
The Mother of Take-Home
Messengers:
THE MOST IMPORTANT THING
IN DATA ANALYSIS, IS NOT THE
DATA ANALYSIS, BUT THE
QUESTION!