Title: Shrunken Centroid Ordering by Orthogonal Projections SCOOP method of variable selection
1Shrunken Centroid Ordering by Orthogonal
Projections(SCOOP) method of variable selection
- Joe Verducci
- Ohio State University
2Outline
- Motivationgene expression
- Variable selection for LDA
- Large p Moderate n
- Advantages in gene selection
- Method
- Model Justification
- Measures of Performance
- Modifications
3LDA Motivation
- Non-greedy selection
- preserve (augmented) discriminant information
- Variables with between group differences
- Variables highly correlated with these
4Fishers Linear Discriminant FunctionandA
Stupid Generalization
where
5Why Its Stupid
S
m1
m2
Results from Bickel and Levina (2004) imply that
the eigenvectors of within and between group
covariance matrices approach orthogonality under
n fixed p?infinity asymptotics.
6Genetic Motivation
- Wound Healing
- 80 National Wound Healing Clinics
- 1000 patients
- Initial 1-week samples
- Clinical records of patients
- 10K genes of potential interest in myocytes
- Subsets of genes act in concert
- A single gene may be active in several subsystems
7P53
- When the DNA in a cell becomes damaged by agents
such as toxic chemicals or ultraviolet (UV) rays
from sunlight, this protein plays a critical role
in determining whether the DNA will be repaired
or the cell will undergo programmed cell death
(apoptosis). - If the DNA can be repaired, tumor protein p53
activates other genes to fix the damage. - If the DNA cannot be repaired, tumor protein p53
prevents the cell from dividing and signals it to
undergo apoptosis. This process prevents cells
with mutated or damaged DNA from dividing, which
helps prevent the development of tumors.
8Pathway construction based on GeneChipTM
expression data. Genes shown in red ellipse are
candidates identified using GeneChipTM assay that
were up-regulated in 20 O2 compared with 3 O2.
Green ellipses are genes that were down-regulated
under conditions mentioned above. The expressions
of candidates shown in red ellipse with blue
outline have been independently verified using
either real-time PCR or ribonuclease protection
assay (6). BAX, Bcl2-associated X protein Catn,
catenin CASP, caspage ccng, cyclin G Cdc61,
cell division cycle CDK, cyclin-dependent
kinase CDKN1A, cyclin-dependent kinase inhibitor
1A (p21) Cx43, gap junction membrane channel
protein GADD, growth arrest and DNA
damage-inducible MAPK, mitogen-activated protein
kinase Mdm2, transformed mouse 3T3 cell double
minute 2 N-Cdh, cadherin 2 PXN, paxillin Tob,
transducer of ErbB-2.1 TP53, transformation-relat
ed protein 53 Vcl, vinculin Wig, wild-type
p53-induced gene 1.
9Motivating Simple Example
- Two groups
- 50 samples in each
- P 4000 normal variables
- All have variance 1
- First 10 variables
- correlation .75 between all pairs
- Difference of 2 between group means
- Second 10 variables
- correlation .75 between all pairs
- Difference of 1 between group means
- Last 3980 variables
- independent
- same mean in both groups
10Results from 100 Simulations
- Individual t-test ranking by p-values
- 73 of top 20 selected are correct
- On average need to select 400 variables to ensure
inclusion of all 20 - SCOOP
- 91 of top 20 selected are correct
- On average need to select 200 variables to ensure
inclusion of all 20
11Shrunken Centroid Methodfor K groupsTibshirani,
Hastie,Narasimhan Chu
- For each gene i,
- xik sample mean in group k,
- xi overall sample mean
- sik estimated std. error of xik
- Based on pooled std deviation
- dik (xik - xi)/sik is a t-statistic
- Shrinking by an amount D gt 0 gives
- Shrunken difference
12Properties of Shrunken Centroid
- When K 2, ordering of variables/genes is same
as t-test - Keeps redundant predictors
- Can be modified to regularize the estimated std
errors - Shrunken centroids used directly for
classification - Shrinkage by amount D is simultaneous in all
coordinates on standardized scale - Shrinkage parameter D chosen by cross-validation
13Reformulating the Goals
- Genetic studies
- Find biomarkers
- classification/prediction
- Use small number of classifiers/predictors
- Understand genetic pathways
- Discover which genes work together to make a
difference - possible intervention
- Other studies
- Improve efficiency in difficult discrimination
problems
14SCOOP Method(version 1)
- Define the Augmented Discriminant Space
- ADS span of eigenvectors
- of Within and Between Covariance Matrices
- Modify shrinkage so as not to distort
configuration of data in the ADS - shrink variables differentially along directions
orthogonal to the ADS - Note Unlike the reference, we do not
standardize, but scale only at the shrinkage
stage. - Keep track of the amount of shrinkage li needed
to eliminate the ith variable
15SCOOP Algorithmfor K groups
- 1. Between Group eigenvectors
- DB (xik - xi) p x K matrix
- Use Singular Value Decompostion (SVD) on DB.
The singular vectors of DB are the eigenvectors
of - DB (DB)T
- 2. Within Group eigenvectors
16Algorithm (part 2)
- Orthogonalize the Between group (BG) eigenvectors
to the Within group (WG) eigenvectors - Note residuals from orthogonalization will no
longer be orthogonal to each other - Renormalize
- compute projection operator onto complement of
the ADS - Note do not need to use p x p storage
17Algorithm (part 3)
- Order variables by scaled shrinkage distances
li - For each variable i, compute a scale value
(squared) length of its projection onto the
orthogonal complement of the ADS - Then calculate how many li such units are
needed to shrink each of the K mean differences
to 0
18Notes
- Shrinking is non-linear
- it truncates at 0
- shrinks each group only as much as it needs to
- What to use as a stopping rule?
- Some measure of preserved information
- Elbow in the distribution of li
- Reference to extreme value distribution
19Theoretical Concern
- Inconsistency of sample eigenvectors
- if p(n)/n ? c gt 0
- Johnstone and Lu (2004)
- Unless sparse representation
- (offset) factor model
- Latent factors account for both
- Correlation among variables
- Group mean differences
20Modeling considerations
- Common offset factor model for gene expression
- latent factors represent biological variation
- random measurement error are uniqueness
components of individual genes. - Normally distributed data
- two populations share the same factor structure
- differ only by the means of the underlying
factors - the restricted maximum likelihood procedure is
the (stupid) generalization of Fishers Linear
Discriminant Analysis (SLDA) that incorporates a
generalized inverse of the pooled sample
covariance matrix. - SLDA seldom works well for real data
- amend overly restrictive assumptions on both
means and covariances.
21More model considerations
- Factors underlying biological variation
- Common factors in 2 groups
- Some with different means in 2 groups
- Some with same mean
- Group specific factors
- Some may have non-zero means
- Some have 0 means
- Unique variation among genes
- Most is noise
- A few of the genes that do not load on any factor
may have different means in the two groups - .
22Model
23Simulation
- n100
- p4000
- G2
- K3
- J(g) 1
- s1
- skF1
- Sj(g)1
- Loadings on common factors
- l1 indicates 1st 10 variables 1
- l2 indicates 2nd 10 variables .55
- l3 indicates 3rd 10 variables 0
- Loadings on Group-specific factors
- L1(1) indicates 4th 10 variables .55
- L1(2) indicates 5th 10 variables 0
- Here is the difference in means
24Shrinkage Needed to Select Top Predictors
25Measures of Performance
- Individual t-test ranking by p-values
- 49 of top 30 selected are correct
- On average need to select 400 variables to
ensure inclusion of all 30 - SCOOP
- 61 of top 30 selected are correct
- On average need to select 200 variables to ensure
inclusion of all 30
26Modifications
- Preserve common and group-distinct within group
sample eigenvectors - Regularize sample eigenvectors using Linear
Perturbation Theory
This is piecewise linear until adjacent
eigenvalues become equal
27Conclusions
- To the extent that something like an offset
factor model holds, incorporating correlations
may substantially improve selection of
discriminating variables (DVs) - Clustering of non-DVs does not seem to have any
serious ill effect - SCOOP is one way to use covariance structure
efficiently
28References
- Bickel PJ and Levina E (2004). Some theory for
Fisher's linear discriminant function, naive
Bayes', and some alternatives when there are many
more variables than observations. Bernoulli 10,
no. 6 9891010. - Tibshirani R, Hastie T,Narasimhan Chu (2002)
Diagnosis of multiple cancer types by shrunken
centroids of gene expression. PNAS 99, no. 10
6567-6572. - Sen, CK, Verducci, JS, Melfi, VF, Khanna, S,
Barbacioru, C and Roy, S (2005). Post-reperfusion
healing of the heart Focus on oxygen-sensitive
genes and DNA microarray as a tool. Mathematical
Biosciences Institute Technical Report No. 31
(available at http//mbi.osu.edu/publications/pub2
005.html)