1.1 Binary file grant.doc has changed

2.1 --- a/grant.html Tue Apr 21 03:36:06 2009 -0700 2.2 +++ b/grant.html Tue Apr 21 04:05:54 2009 -0700 2.3 @@ -592,16 +592,14 @@ 2.4 _____________________________ 2.5 195-fold cross-validation. 2.6 Research Design and Methods 2.7 -Further work on flatmapping 2.8 -Often the surface of a structure serves as a natural 2-D basis for anatomical organization. Even when the shape of the 2.9 -surface is known, there are multiple ways to map it into a plane. We will compare mappings which attempt to preserve 2.10 -size (such as the one used by Caret[7]) with mappings which preserve angle (conformal maps). Although there is much 2-D 2.11 -organization in anatomy, there are also structures whose anatomy is fundamentally 3-dimensional. We plan to include a 2.12 -statistical test that warns the user if the assumption of 2-D structure seems to be wrong. 2.13 -Automatic segmentation of cortical layers 2.14 -Extension to probabalistic maps Presently, we do not have a probabalistic atlas which is registered to the ABA 2.15 -space. However, in anticipation of the availability of such maps, we would like to explore extensions to our Aim 1 techniques 2.16 -which can handle probabalistic maps. 2.17 +Flatmapping and segmentation of cortical layers** 2.18 +There are multiple ways to flatten 3-D data into 2-D. We will compare mappings from manifolds to planes which attempt 2.19 +to preserve size (such as the one used by Caret[7]) with mappings which preserve angle (conformal maps). Our method will 2.20 +include a statistical test that warns the user if the assumption of 2-D structure seems to be wrong. 2.21 +We have not yet made use of radial profiles. While the radial profiles may be used “raw”, for laminar structures like the 2.22 +cortex another strategy is to group together voxels in the same cortical layer; each surface pixel would then be associated 2.23 +with one expression level per gene per layer. We will develop a segmentation algorithm to automatically identify the layer 2.24 +boundaries. 2.25 Develop algorithms that find genetic markers for anatomical regions 2.26 1.Develop scoring measures for evaluating how good individual genes are at marking areas: we will compare pointwise, 2.27 geometric, and information-theoretic measures. 2.28 @@ -620,33 +618,30 @@ 2.29 # Linear discriminant analysis 2.30 Decision trees todo 2.31 20. 2.32 -Apply these algorithms to the cortex 2.33 -1.Create open source format conversion tools: we will create tools to bulk download the ABA dataset and to convert 2.34 -between SEV, NIFTI and MATLAB formats. 2.35 -2.Flatmap the ABA cortex data: map the ABA data onto a plane and draw the cortical area boundaries onto it. 2.36 -3.Find layer boundaries: cluster similar voxels together in order to automatically find the cortical layer boundaries. 2.37 -4.Run the procedures that we developed on the cortex: we will present, for each area, a short list of markers to identify 2.38 -that area; and we will also present lists of “panels” of genes that can be used to delineate many areas at once. 2.39 +# confirm with EMAGE, GeneAtlas, GENSAT, etc, to fight overfitting, two hemis 2.40 +# mixture models, etc 2.41 Develop algorithms to suggest a division of a structure into anatomical parts 2.42 -# mixture models, etc 2.43 1.Explore dimensionality reduction algorithms applied to pixels: including TODO 2.44 2.Explore dimensionality reduction algorithms applied to genes: including TODO 2.45 3.Explore clustering algorithms applied to pixels: including TODO 2.46 -_________________________________________ 2.47 - 20Already, for each cortical area, we have used the C4.5 algorithm to find a decision tree for that area. We achieved good classification accuracy 2.48 -on our training set, but the number of genes that appeared in each tree was too large. We plan to implement a pruning procedure to generate 2.49 -trees that use fewer genes 2.50 4.Explore clustering algorithms applied to genes: including gene shaving, TODO 2.51 5.Develop an algorithm to use dimensionality reduction and/or hierarchial clustering to create anatomical maps 2.52 6.Run this algorithm on the cortex: present a hierarchial, genoarchitectonic map of the cortex 2.53 # Linear discriminant analysis 2.54 # jbt, coclustering 2.55 # self-organizing map 2.56 -# confirm with EMAGE, GeneAtlas, GENSAT, etc, to fight overfitting, two hemis 2.57 # compare using clustering scores 2.58 # multivariate gradient similarity 2.59 # deep belief nets 2.60 -# note: slice artifact 2.61 +Apply these algorithms to the cortex Using the methods developed in Aim 1, we will present, for each cortical area, 2.62 +a short list of markers to identify that area; and we will also present lists of “panels” of genes that can be used to delineate 2.63 +_________________________________________ 2.64 + 20Already, for each cortical area, we have used the C4.5 algorithm to find a decision tree for that area. We achieved good classification accuracy 2.65 +on our training set, but the number of genes that appeared in each tree was too large. We plan to implement a pruning procedure to generate 2.66 +trees that use fewer genes 2.67 +many areas at once. Using the methods developed in Aim 2, we will present one or more hierarchial cortical maps. We will 2.68 +identifyand explain how the statistical structure in the gene expression data led to any unexpected or interesting features 2.69 +of thesemaps. 2.70 Bibliography & References Cited 2.71 [1]Chris Adamson, Leigh Johnston, Terrie Inder, Sandra Rees, Iven Mareels, and Gary Egan. A Tracking Approach to 2.72 Parcellation of the Cerebral Cortex, volume Volume 3749/2005 of Lecture Notes in Computer Science, pages 294–301.

3.1 Binary file grant.odt has changed

4.1 Binary file grant.pdf has changed

5.1 --- a/grant.txt Tue Apr 21 03:36:06 2009 -0700 5.2 +++ b/grant.txt Tue Apr 21 04:05:54 2009 -0700 5.3 @@ -448,26 +448,27 @@ 5.4 == Research Design and Methods == 5.5 5.6 5.7 -\vspace{0.3cm}**Further work on flatmapping** 5.8 +\vspace{0.3cm}**Flatmapping and segmentation of cortical layers** 5.9 5.10 %%In anatomy, the manifold of interest is usually either defined by a combination of two relevant anatomical axes (todo), or by the surface of the structure (as is the case with the cortex). In the former case, the manifold of interest is a plane, but in the latter case it is curved. If the manifold is curved, there are various methods for mapping the manifold into a plane. 5.11 5.12 %%In the case of the cerebral cortex, it remains to be seen which method of mapping the manifold into a plane is optimal for this application. We will compare mappings which attempt to preserve size (such as the one used by Caret\cite{van_essen_integrated_2001}) with mappings which preserve angle (conformal maps). 5.13 5.14 5.15 -Often the surface of a structure serves as a natural 2-D basis for anatomical organization. Even when the shape of the surface is known, there are multiple ways to map it into a plane. We will compare mappings which attempt to preserve size (such as the one used by Caret\cite{van_essen_integrated_2001}) with mappings which preserve angle (conformal maps). Although there is much 2-D organization in anatomy, there are also structures whose anatomy is fundamentally 3-dimensional. We plan to include a statistical test that warns the user if the assumption of 2-D structure seems to be wrong. 5.16 - 5.17 -\vspace{0.3cm}**Automatic segmentation of cortical layers** 5.18 - 5.19 - 5.20 - 5.21 -\vspace{0.3cm}**Extension to probabalistic maps** 5.22 -Presently, we do not have a probabalistic atlas which is registered to the ABA space. However, in anticipation of the availability of such maps, we would like to explore extensions to our Aim 1 techniques which can handle probabalistic maps. 5.23 - 5.24 - 5.25 +%%Often the surface of a structure serves as a natural 2-D basis for anatomical organization. Even when the shape of the surface is known, there are multiple ways to map it into a plane. We will compare mappings which attempt to preserve size (such as the one used by Caret\cite{van_essen_integrated_2001}) with mappings which preserve angle (conformal maps). Although there is much 2-D organization in anatomy, there are also structures whose anatomy is fundamentally 3-dimensional. We plan to include a statistical test that warns the user if the assumption of 2-D structure seems to be wrong. 5.26 + 5.27 +There are multiple ways to flatten 3-D data into 2-D. We will compare mappings from manifolds to planes which attempt to preserve size (such as the one used by Caret\cite{van_essen_integrated_2001}) with mappings which preserve angle (conformal maps). Our method will include a statistical test that warns the user if the assumption of 2-D structure seems to be wrong. 5.28 + 5.29 +We have not yet made use of radial profiles. While the radial profiles may be used "raw", for laminar structures like the cortex another strategy is to group together voxels in the same cortical layer; each surface pixel would then be associated with one expression level per gene per layer. We will develop a segmentation algorithm to automatically identify the layer boundaries. 5.30 5.31 \vspace{0.3cm}**Develop algorithms that find genetic markers for anatomical regions** 5.32 5.33 + 5.34 + 5.35 + 5.36 + 5.37 + 5.38 + 5.39 # Develop scoring measures for evaluating how good individual genes are at marking areas: we will compare pointwise, geometric, and information-theoretic measures. 5.40 # Develop a procedure to find single marker genes for anatomical regions: for each cortical area, by using or combining the scoring measures developed, we will rank the genes by their ability to delineate each area. 5.41 # Extend the procedure to handle difficult areas by using combinatorial coding: for areas that cannot be identified by any single gene, identify them with a handful of genes. We will consider both (a) algorithms that incrementally/greedily combine single gene markers into sets, such as forward stepwise regression and decision trees, and also (b) supervised learning techniques which use soft constraints to minimize the number of features, such as sparse support vector machines. 5.42 @@ -481,20 +482,14 @@ 5.43 5.44 \footnote{Already, for each cortical area, we have used the C4.5 algorithm to find a decision tree for that area. We achieved good classification accuracy on our training set, but the number of genes that appeared in each tree was too large. We plan to implement a pruning procedure to generate trees that use fewer genes}. 5.45 5.46 - 5.47 -\vspace{0.3cm}**Apply these algorithms to the cortex** 5.48 - 5.49 -# Create open source format conversion tools: we will create tools to bulk download the ABA dataset and to convert between SEV, NIFTI and MATLAB formats. 5.50 -# Flatmap the ABA cortex data: map the ABA data onto a plane and draw the cortical area boundaries onto it. 5.51 -# Find layer boundaries: cluster similar voxels together in order to automatically find the cortical layer boundaries. 5.52 -# Run the procedures that we developed on the cortex: we will present, for each area, a short list of markers to identify that area; and we will also present lists of "panels" of genes that can be used to delineate many areas at once. 5.53 +# confirm with EMAGE, GeneAtlas, GENSAT, etc, to fight overfitting, two hemis 5.54 + 5.55 +# mixture models, etc 5.56 5.57 5.58 5.59 \vspace{0.3cm}**Develop algorithms to suggest a division of a structure into anatomical parts** 5.60 5.61 -# mixture models, etc 5.62 - 5.63 # Explore dimensionality reduction algorithms applied to pixels: including TODO 5.64 # Explore dimensionality reduction algorithms applied to genes: including TODO 5.65 # Explore clustering algorithms applied to pixels: including TODO 5.66 @@ -508,7 +503,6 @@ 5.67 5.68 # self-organizing map 5.69 5.70 -# confirm with EMAGE, GeneAtlas, GENSAT, etc, to fight overfitting, two hemis 5.71 5.72 5.73 # compare using clustering scores 5.74 @@ -517,7 +511,17 @@ 5.75 5.76 # deep belief nets 5.77 5.78 -# note: slice artifact 5.79 + 5.80 + 5.81 +\vspace{0.3cm}**Apply these algorithms to the cortex** 5.82 +Using the methods developed in Aim 1, we will present, for each cortical area, a short list of markers to identify that area; and we will also present lists of "panels" of genes that can be used to delineate many areas at once. Using the methods developed in Aim 2, we will present one or more hierarchial cortical maps. We will identify and explain how the statistical structure in the gene expression data led to any unexpected or interesting features of these maps. 5.83 + 5.84 + 5.85 +%%# note: slice artifact 5.86 + 5.87 +%%\vspace{0.3cm}**Extension to probabalistic maps** 5.88 +%%Presently, we do not have a probabalistic atlas which is registered to the ABA space. However, in anticipation of the availability of such maps, we would like to explore extensions to our Aim 1 techniques which can handle probabalistic maps. 5.89 + 5.90 5.91 \newpage 5.92