Transcript FatiScan

::: Schedule.
1.
2.
3.
4.
Biological (Functional) Databases
Threshold-based and threshold free methods
Threshold-based example: FatiGO.
Threshold free example 1: FatisScan.
Two-steps approach reproduces pre-genomics paradigms
experiments
experiments
pass
test
test
test
test
test
test
test
test
test
....
annotation
no
annotation
:
:
Context and cooperation between genes is ignored
A previous step of gene selection causes loss of information and
makes the test insensitive
Significantly
over-expressed
in B
p<0.05
statistic
t-test
with two
tails.
-
AB
GO1
GO2
Very few genes selected to
arrive to a significant
conclussion on GO1 and GO2
If a threshold based on the
experimental values is applied,
and the resulting selection of
genes compared for
enrichment of a functional
term, this might not be found
Functional Classes expressed
as blocks in A and B
Significantly
over-expressed
in A
+
Threshold-free approach
Including information in the procedure of gene selection
Functional label A
List of genes
+
A
B
C
Functional label B
Blocks of genes
with significant
functional term
A annotated
•Genes are ranked by
any biological criteria
(expression value, pvalue, positive
selection, etc).
Homogeneously
distributed
Functional term
C
•FatiScan searchs for
the distribution of the
blocks of functionally
related genes across
the list.
Functional label C
•To detect significant
terms a segmentation
test is performed.
-
Blocks of genes
with significant
functional term
B annotated
•Properties of groups
of genes are
considered.
Gene Set Enrichment
Just ONE list
(ranked)
::: Exercise 3: FatiScan
Selected data
Mootha et al.
Nat Genet 2003
-RunT-rex for differential expression analysisFiles:
fatiscan_diabetes_mootha_array.txt
fatiscan_diabetes_mootha_classes.txt
Results
T-Rex
Run
T-Rex
here
http://www.gepas.org/
-Examine Results and send to FatiScan-
Send to FatiScan here
Check the FDR
adjusted p-value
-Run Fatiscan with statistic-ranked list
In FatiScan input form: select the
organism Homo sapiens, Gene
Ontology: biological process with
the default filtering, 30 partitions,
Two tailed Fisher's exact test, and
press Run. As the sorted list comes
from GEPAS it will automatically
detect the sorting and the labels.
Significantly altered KEGG Pathways (MGT y DM2/IGT):