cg
changeset 75:e6f8c1f1539f
.
author | bshanks@bshanks.dyndns.org |
---|---|
date | Mon Apr 20 17:09:48 2009 -0700 (16 years ago) |
parents | 19647ba662c8 |
children | b262a04b6a66 |
files | grant.doc grant.html grant.pdf grant.txt |
line diff
1.1 Binary file grant.doc has changed
2.1 --- a/grant.html Mon Apr 20 17:02:27 2009 -0700
2.2 +++ b/grant.html Mon Apr 20 17:09:48 2009 -0700
2.3 @@ -319,33 +319,9 @@
2.4 most cortical areas is that in Gene Finder, although the user chooses a seed voxel, Gene Finder chooses the ROI for which genes will be found,
2.5 and it creates that ROI by (pairwise voxel correlation) clustering around the seed.
2.6 Preliminary work
2.7 -Format conversion between SEV, MATLAB, NIFTI
2.8 -We have created software to (politely) download all of the SEV files from the Allen Institute website. We have also created
2.9 -software to convert between the SEV, MATLAB, and NIFTI file formats, as well as some of Caret’s file formats.
2.10 -Flatmap of cortex
2.11 -
2.12 -Figure 1: Gene Pitx2
2.13 -is selectively underex-
2.14 -pressed in area SS (so-
2.15 -matosensory). We downloaded the ABA data and applied a mask to select only those voxels which belong to
2.16 - cerebral cortex. We divided the cortex into hemispheres.
2.17 - Using Caret[5], we created a mesh representation of the surface of the selected voxels. For
2.18 - each gene, for each node of the mesh, we calculated an average of the gene expression of the
2.19 - voxels “underneath” that mesh node. We then flattened the cortex, creating a two-dimensional
2.20 - mesh.
2.21 - We sampled the nodes of the irregular, flat mesh in order to create a regular grid of pixel
2.22 - values. We converted this grid into a MATLAB matrix.
2.23 - We manually traced the boundaries of each of 49 cortical areas from the ABA coronal reference
2.24 - atlas slides. We then converted these manual traces into Caret-format regional boundary data
2.25 - on the mesh surface. We projected the regions onto the 2-d mesh, and then onto the grid, and
2.26 - then we converted the region data into MATLAB format.
2.27 - At this point, the data is in the form of a number of 2-D matrices, all in registration, with
2.28 - the matrix entries representing a grid of points (pixels) over the cortical surface:
2.29 -∙A 2-D matrix whose entries represent the regional label associated with each surface pixel
2.30 -∙For each gene, a 2-D matrix whose entries represent the average expression level underneath each surface pixel
2.31 -
2.32 -
2.33 -Figure 2: Top row: Genes Nfic and
2.34 +
2.35 +
2.36 +Figure 1: Top row: Genes Nfic and
2.37 A930001M12Rik are the most correlated with
2.38 area SS (somatosensory cortex). Bottom row:
2.39 Genes C130038G02Rik and Cacna1i are those
2.40 @@ -357,37 +333,61 @@
2.41 at the right. The red outline is the boundary
2.42 of region MO. Pixels are colored according to
2.43 correlation, with red meaning high correlation
2.44 -and blue meaning low. We created a normalized version of the gene expression data by sub-
2.45 - tracting each gene’s mean expression level (over all surface pixels) and
2.46 - dividing each gene by its standard deviation.
2.47 - The features and the target area are both functions on the surface
2.48 - pixels. They can be referred to as scalar fields over the space of sur-
2.49 - face pixels; alternately, they can be thought of as images which can be
2.50 - displayed on the flatmapped surface.
2.51 - To move beyond a single average expression level for each surface
2.52 - pixel, we plan to create a separate matrix for each cortical layer to rep-
2.53 - resent the average expression level within that layer. Cortical layers are
2.54 - found at different depths in different parts of the cortex. In preparation
2.55 - for extracting the layer-specific datasets, we have extended Caret with
2.56 - routines that allow the depth of the ROI for volume-to-surface projection
2.57 - to vary.
2.58 - In the Research Plan, we describe how we will automatically locate
2.59 - the layer depths. For validation, we have manually demarcated the depth
2.60 - of the outer boundary of cortical layer 5 throughout the cortex.
2.61 - Feature selection and scoring methods
2.62 - Underexpression of a gene can serve as a marker Underexpression
2.63 - of a gene can sometimes serve as a marker. See, for example, Figure 1.
2.64 - Correlation Recall that the instances are surface pixels, and con-
2.65 - sider the problem of attempting to classify each instance as either a
2.66 - member of a particular anatomical area, or not. The target area can be
2.67 - represented as a boolean mask over the surface pixels.
2.68 - One class of feature selection scoring method are those which calcu-
2.69 - late some sort of “match” between each gene image and the target image.
2.70 - Those genes which match the best are good candidates for features.
2.71 - One of the simplest methods in this class is to use correlation as
2.72 - the match score. We calculated the correlation between each gene and
2.73 - each cortical area. The top row of Figure 2 shows the three genes most
2.74 -correlated with area SS.
2.75 +and blue meaning low. Format conversion between SEV, MATLAB, NIFTI
2.76 + We have created software to (politely) download all of the SEV files from
2.77 + the Allen Institute website. We have also created software to convert
2.78 + between the SEV, MATLAB, and NIFTI file formats, as well as some of
2.79 + Caret’s file formats.
2.80 + Flatmap of cortex
2.81 + We downloaded the ABA data and applied a mask to select only those
2.82 + voxels which belong to cerebral cortex. We divided the cortex into hemi-
2.83 + spheres.
2.84 + Using Caret[5], we created a mesh representation of the surface of the
2.85 + selected voxels. For each gene, for each node of the mesh, we calculated
2.86 + an average of the gene expression of the voxels “underneath” that mesh
2.87 + node. We then flattened the cortex, creating a two-dimensional mesh.
2.88 + We sampled the nodes of the irregular, flat mesh in order to create
2.89 + a regular grid of pixel values. We converted this grid into a MATLAB
2.90 + matrix.
2.91 + We manually traced the boundaries of each of 49 cortical areas from
2.92 + the ABA coronal reference atlas slides. We then converted these manual
2.93 + traces into Caret-format regional boundary data on the mesh surface.
2.94 + We projected the regions onto the 2-d mesh, and then onto the grid, and
2.95 + then we converted the region data into MATLAB format.
2.96 + At this point, the data is in the form of a number of 2-D matrices,
2.97 + all in registration, with the matrix entries representing a grid of points
2.98 + (pixels) over the cortical surface:
2.99 + ∙ A 2-D matrix whose entries represent the regional label associated with
2.100 + each surface pixel
2.101 + ∙ For each gene, a 2-D matrix whose entries represent the average expres-
2.102 + sion level underneath each surface pixel
2.103 +
2.104 +Figure 2: Gene Pitx2
2.105 +is selectively underex-
2.106 +pressed in area SS (so-
2.107 +matosensory). We created a normalized version of the gene expression data by subtracting each gene’s mean
2.108 + expression level (over all surface pixels) and dividing each gene by its standard deviation.
2.109 + The features and the target area are both functions on the surface pixels. They can be referred
2.110 + to as scalar fields over the space of surface pixels; alternately, they can be thought of as images
2.111 + which can be displayed on the flatmapped surface.
2.112 + To move beyond a single average expression level for each surface pixel, we plan to create a
2.113 + separate matrix for each cortical layer to represent the average expression level within that layer.
2.114 + Cortical layers are found at different depths in different parts of the cortex. In preparation for
2.115 + extracting the layer-specific datasets, we have extended Caret with routines that allow the depth
2.116 + of the ROI for volume-to-surface projection to vary.
2.117 + In the Research Plan, we describe how we will automatically locate the layer depths. For
2.118 + validation, we have manually demarcated the depth of the outer boundary of cortical layer 5
2.119 + throughout the cortex.
2.120 + Feature selection and scoring methods
2.121 +Underexpression of a gene can serve as a marker Underexpression of a gene can sometimes serve as a marker. See,
2.122 +for example, Figure 2.
2.123 +Correlation Recall that the instances are surface pixels, and consider the problem of attempting to classify each instance
2.124 +as either a member of a particular anatomical area, or not. The target area can be represented as a boolean mask over the
2.125 +surface pixels.
2.126 +One class of feature selection scoring method are those which calculate some sort of “match” between each gene image
2.127 +and the target image. Those genes which match the best are good candidates for features.
2.128 +One of the simplest methods in this class is to use correlation as the match score. We calculated the correlation between
2.129 +each gene and each cortical area. The top row of Figure 1 shows the three genes most correlated with area SS.
2.130
2.131
2.132 Figure 3: The top row shows the two genes which
3.1 Binary file grant.pdf has changed
4.1 --- a/grant.txt Mon Apr 20 17:02:27 2009 -0700
4.2 +++ b/grant.txt Mon Apr 20 17:09:48 2009 -0700
4.3 @@ -228,12 +228,42 @@
4.4 \newpage
4.5
4.6 == Preliminary work ==
4.7 +\begin{wrapfigure}{L}{0.4\textwidth}\centering
4.8 +%%\includegraphics[scale=.31]{singlegene_SS_corr_top_1_2365_jet.eps}\includegraphics[scale=.31]{singlegene_SS_corr_top_2_242_jet.eps}\includegraphics[scale=.31]{singlegene_SS_corr_top_3_654_jet.eps}
4.9 +%%\\
4.10 +%%\includegraphics[scale=.31]{singlegene_SS_lr_top_1_654_jet.eps}\includegraphics[scale=.31]{singlegene_SS_lr_top_2_685_jet.eps}\includegraphics[scale=.31]{singlegene_SS_lr_top_3_724_jet.eps}
4.11 +%%\caption{Top row: Genes Nfic, A930001M12Rik, C130038G02Rik are the most correlated with area SS (somatosensory cortex). Bottom row: Genes C130038G02Rik, Cacna1i, Car10 are those with the best fit using logistic regression. Within each picture, the vertical axis roughly corresponds to anterior at the top and posterior at the bottom, and the horizontal axis roughly corresponds to medial at the left and lateral at the right. The red outline is the boundary of region SS. Pixels are colored according to correlation, with red meaning high correlation and blue meaning low.}
4.12 +
4.13 +\includegraphics[scale=.31]{singlegene_SS_corr_top_1_2365_jet.eps}\includegraphics[scale=.31]{singlegene_SS_corr_top_2_242_jet.eps}
4.14 +\\
4.15 +\includegraphics[scale=.31]{singlegene_SS_lr_top_1_654_jet.eps}\includegraphics[scale=.31]{singlegene_SS_lr_top_2_685_jet.eps}
4.16 +
4.17 +\caption{Top row: Genes Nfic and A930001M12Rik are the most correlated with area SS (somatosensory cortex). Bottom row: Genes C130038G02Rik and Cacna1i are those with the best fit using logistic regression. Within each picture, the vertical axis roughly corresponds to anterior at the top and posterior at the bottom, and the horizontal axis roughly corresponds to medial at the left and lateral at the right. The red outline is the boundary of region SS. Pixels are colored according to correlation, with red meaning high correlation and blue meaning low.}
4.18 +\label{SScorrLr}\end{wrapfigure}
4.19 +
4.20 +
4.21 +
4.22
4.23 === Format conversion between SEV, MATLAB, NIFTI ===
4.24 We have created software to (politely) download all of the SEV files from the Allen Institute website. We have also created software to convert between the SEV, MATLAB, and NIFTI file formats, as well as some of Caret's file formats.
4.25
4.26
4.27 === Flatmap of cortex ===
4.28 +
4.29 +
4.30 +We downloaded the ABA data and applied a mask to select only those voxels which belong to cerebral cortex. We divided the cortex into hemispheres.
4.31 +
4.32 +Using Caret\cite{van_essen_integrated_2001}, we created a mesh representation of the surface of the selected voxels. For each gene, for each node of the mesh, we calculated an average of the gene expression of the voxels "underneath" that mesh node. We then flattened the cortex, creating a two-dimensional mesh.
4.33 +
4.34 +We sampled the nodes of the irregular, flat mesh in order to create a regular grid of pixel values. We converted this grid into a MATLAB matrix.
4.35 +
4.36 +We manually traced the boundaries of each of 49 cortical areas from the ABA coronal reference atlas slides. We then converted these manual traces into Caret-format regional boundary data on the mesh surface. We projected the regions onto the 2-d mesh, and then onto the grid, and then we converted the region data into MATLAB format.
4.37 +
4.38 +At this point, the data is in the form of a number of 2-D matrices, all in registration, with the matrix entries representing a grid of points (pixels) over the cortical surface:
4.39 +
4.40 +* A 2-D matrix whose entries represent the regional label associated with each surface pixel
4.41 +* For each gene, a 2-D matrix whose entries represent the average expression level underneath each surface pixel
4.42 +
4.43 \begin{wrapfigure}{L}{0.2\textwidth}\centering
4.44 \includegraphics[scale=.31]{holeExample_2682_SS_jet.eps}
4.45 \caption{Gene Pitx2 is selectively underexpressed in area SS (somatosensory).}
4.46 @@ -241,34 +271,6 @@
4.47
4.48
4.49
4.50 -We downloaded the ABA data and applied a mask to select only those voxels which belong to cerebral cortex. We divided the cortex into hemispheres.
4.51 -
4.52 -Using Caret\cite{van_essen_integrated_2001}, we created a mesh representation of the surface of the selected voxels. For each gene, for each node of the mesh, we calculated an average of the gene expression of the voxels "underneath" that mesh node. We then flattened the cortex, creating a two-dimensional mesh.
4.53 -
4.54 -We sampled the nodes of the irregular, flat mesh in order to create a regular grid of pixel values. We converted this grid into a MATLAB matrix.
4.55 -
4.56 -We manually traced the boundaries of each of 49 cortical areas from the ABA coronal reference atlas slides. We then converted these manual traces into Caret-format regional boundary data on the mesh surface. We projected the regions onto the 2-d mesh, and then onto the grid, and then we converted the region data into MATLAB format.
4.57 -
4.58 -At this point, the data is in the form of a number of 2-D matrices, all in registration, with the matrix entries representing a grid of points (pixels) over the cortical surface:
4.59 -
4.60 -* A 2-D matrix whose entries represent the regional label associated with each surface pixel
4.61 -* For each gene, a 2-D matrix whose entries represent the average expression level underneath each surface pixel
4.62 -
4.63 -\begin{wrapfigure}{L}{0.4\textwidth}\centering
4.64 -%%\includegraphics[scale=.31]{singlegene_SS_corr_top_1_2365_jet.eps}\includegraphics[scale=.31]{singlegene_SS_corr_top_2_242_jet.eps}\includegraphics[scale=.31]{singlegene_SS_corr_top_3_654_jet.eps}
4.65 -%%\\
4.66 -%%\includegraphics[scale=.31]{singlegene_SS_lr_top_1_654_jet.eps}\includegraphics[scale=.31]{singlegene_SS_lr_top_2_685_jet.eps}\includegraphics[scale=.31]{singlegene_SS_lr_top_3_724_jet.eps}
4.67 -%%\caption{Top row: Genes Nfic, A930001M12Rik, C130038G02Rik are the most correlated with area SS (somatosensory cortex). Bottom row: Genes C130038G02Rik, Cacna1i, Car10 are those with the best fit using logistic regression. Within each picture, the vertical axis roughly corresponds to anterior at the top and posterior at the bottom, and the horizontal axis roughly corresponds to medial at the left and lateral at the right. The red outline is the boundary of region MO. Pixels are colored according to correlation, with red meaning high correlation and blue meaning low.}
4.68 -
4.69 -\includegraphics[scale=.31]{singlegene_SS_corr_top_1_2365_jet.eps}\includegraphics[scale=.31]{singlegene_SS_corr_top_2_242_jet.eps}
4.70 -\\
4.71 -\includegraphics[scale=.31]{singlegene_SS_lr_top_1_654_jet.eps}\includegraphics[scale=.31]{singlegene_SS_lr_top_2_685_jet.eps}
4.72 -
4.73 -\caption{Top row: Genes Nfic and A930001M12Rik are the most correlated with area SS (somatosensory cortex). Bottom row: Genes C130038G02Rik and Cacna1i are those with the best fit using logistic regression. Within each picture, the vertical axis roughly corresponds to anterior at the top and posterior at the bottom, and the horizontal axis roughly corresponds to medial at the left and lateral at the right. The red outline is the boundary of region MO. Pixels are colored according to correlation, with red meaning high correlation and blue meaning low.}
4.74 -\label{SScorrLr}\end{wrapfigure}
4.75 -
4.76 -
4.77 -
4.78 We created a normalized version of the gene expression data by subtracting each gene's mean expression level (over all surface pixels) and dividing each gene by its standard deviation.
4.79
4.80 The features and the target area are both functions on the surface pixels. They can be referred to as scalar fields over the space of surface pixels; alternately, they can be thought of as images which can be displayed on the flatmapped surface.