cg

changeset 44:c4a887af9b0b

.
author bshanks@bshanks.dyndns.org
date Wed Apr 15 03:19:01 2009 -0700 (16 years ago)
parents 8cce366da1e5
children 354ea5edb5f6
files grant.html grant.odt grant.pdf grant.txt
line diff
1.1 --- a/grant.html Wed Apr 15 00:50:34 2009 -0700 1.2 +++ b/grant.html Wed Apr 15 03:19:01 2009 -0700 1.3 @@ -82,22 +82,22 @@ 1.4 data. 1.5 Therefore, when possible, the instances should represent pixels, not voxels. 1.6 Related work 1.7 -There is a substantial body of work on the analysis of gene expression data, however, most of this concerns gene expression 1.8 -data which is not fundamentally spatial. 1.9 +There is a substantial body of work on the analysis of gene expression data, most of this concerns gene expression data 1.10 +which is not fundamentally spatial2. 1.11 As noted above, there has been much work on both supervised learning and there are many available algorithms for 1.12 each. However, the algorithms require the scientist to provide a framework for representing the problem domain, and the 1.13 way that this framework is set up has a large impact on performance. Creating a good framework can require creatively 1.14 reconceptualizing the problem domain, and is not merely a mechanical “fine-tuning” of numerical parameters. For example, 1.15 we believe that domain-specific scoring measures (such as gradient similarity, which is discussed in Preliminary Work) may 1.16 be necessary in order to achieve the best results in this application. 1.17 -We are aware of three existing efforts to find marker genes using spatial gene expression data using automated methods. 1.18 -[? ] describes GeneAtlas. GeneAtlas allows the user to construct a search query by freely demarcating one or two 2-D 1.19 +We are aware of four existing efforts to find marker genes using spatial gene expression data using automated methods. 1.20 +[1 ] describes GeneAtlas. GeneAtlas allows the user to construct a search query by freely demarcating one or two 2-D 1.21 regions on sagittal slices, and then to specify either the strength of expression or the name of another gene whose expression 1.22 pattern is to be matched. GeneAtlas differs from our Aim 1 in at least two ways. First, GeneAtlas finds only single genes, 1.23 -whereas we will also look for combinations of genes2. Second, at least for the custom spatial search, Gene Atlas appears to 1.24 +whereas we will also look for combinations of genes3. Second, at least for the custom spatial search, Gene Atlas appears to 1.25 use a simple pointwise scoring method (strength of expression), whereas we will also use geometric metrics such as gradient 1.26 similarity. 1.27 -[2 ] describes AGEA, ”Anatomic Gene Expression Atlas”. AGEA has three components: 1.28 +[6 ] describes AGEA, ”Anatomic Gene Expression Atlas”. AGEA has three components: 1.29 * Gene Finder: The user selects a seed voxel and the system (1) chooses a cluster which includes the seed voxel, (2) 1.30 yields a list of genes which are overexpressed in that cluster. (note: the ABA website also contains pre-prepared lists of 1.31 overexpressed genes for selected structures) 1.32 @@ -107,8 +107,10 @@ 1.33 with correlation as the similarity metric. 1.34 Gene Finder is different from our Aim 1 in at least three ways. First, Gene Finder finds only single genes, whereas we 1.35 will also look for combinations of genes. Second, gene finder can only use overexpression as a marker, whereas we will also 1.36 -search for underexpression. Third, Gene Finder uses a simple pointwise score3, whereas we will also use geometric scores 1.37 +search for underexpression. Third, Gene Finder uses a simple pointwise score4, whereas we will also use geometric scores 1.38 such as gradient similarity. The Preliminary Data section contains evidence that each of our three choices is the right one. 1.39 +[10 ] todo 1.40 +[4 ] todo 1.41 In summary, none of the previous projects explores combinations of marker genes, and none of their publications compare 1.42 the results obtained by using different algorithms or scoring methods. 1.43 Aim 2 1.44 @@ -119,15 +121,17 @@ 1.45 clustering or cluster analysis. 1.46 The task of deciding how to carve up a structure into anatomical regions can be put into these terms. The instances are 1.47 once again voxels (or pixels) along with their associated gene expression profiles. We make the assumption that voxels from 1.48 +_________________________________________ 1.49 + 2By “fundamentally spatial” we mean that there is information from a large number of spatial locations; not just data which has only a few 1.50 +different locations. 1.51 + 3See Preliminary Data for an example of an area which cannot be marked by any single gene in the dataset, but which can be marked by a 1.52 +combination. 1.53 + 4“Expression energy ratio”, which captures overexpression. 1.54 the same region have similar gene expression profiles, at least compared to the other regions. This means that clustering 1.55 voxels is the same as finding potential regions; we seek a partitioning of the voxels into regions, that is, into clusters of voxels 1.56 with similar gene expression. 1.57 -_________________ 1.58 - 2See Preliminary Data for an example of an area which cannot be marked by any single gene in the dataset, but which can be marked by a 1.59 -combination. 1.60 - 3“Expression energy ratio”, which captures overexpression. 1.61 It is desirable to determine not just one set of regions, but also how these regions relate to each other, if at all; perhaps 1.62 -some ofthe regions are more similar to each other than to the rest, suggesting that, although at a fine spatial scale they 1.63 +some of the regions are more similar to each other than to the rest, suggesting that, although at a fine spatial scale they 1.64 could be considered separate, on a coarser spatial scale they could be grouped together into one large region. This suggests 1.65 the outcome of clustering may be a hierarchial tree of clusters, rather than a single set of clusters which partition the voxels. 1.66 This is called hierarchial clustering. 1.67 @@ -172,30 +176,28 @@ 1.68 have one reduced feature for each gene cluster. 1.69 Gene clusters could also be used to directly yield a clustering on instances. This is because many genes have an expression 1.70 pattern which seems to pick out a single, spatially continguous region. Therefore, it seems likely that an anatomically 1.71 -interesting region will have multiple genes which each individually pick it out4. This suggests the following procedure: 1.72 +interesting region will have multiple genes which each individually pick it out5. This suggests the following procedure: 1.73 cluster together genes which pick out similar regions, and then to use the more popular common regions as the final clusters. 1.74 In the Preliminary Data we show that a number of anatomically recognized cortical regions, as well as some “superregions” 1.75 formed by lumping together a few regions, are associated with gene clusters in this fashion. 1.76 +_________________________________________ 1.77 + 5This would seem to contradict our finding in aim 1 that some cortical areas are combinatorially coded by multiple genes. However, it is 1.78 +possible that the currently accepted cortical maps divide the cortex into regions which are unnatural from the point of view of gene expression; 1.79 +perhaps there is some other way to map the cortex for which each region can be identified by single genes. Another possibility is that, although 1.80 +the cluster prototype fits an anatomical region, the individual genes are each somewhat different from the prototype. 1.81 Related work 1.82 We are aware of three existing efforts to cluster spatial gene expression data. 1.83 -_________________________________________ 1.84 - 4This would seem to contradict our finding in aim 1 that some cortical areas are combinatorially coded by multiple genes. However, it is 1.85 -possible that the currently accepted cortical maps divide the cortex into regions which are unnatural from the point of view of gene expression; 1.86 -perhaps there is some other way to map the cortex for which each region can be identified by single genes. 1.87 -[5 ] describes an analysis of the anatomy of the hippocampus using the ABA dataset. In addition to manual analysis, 1.88 +[9 ] describes an analysis of the anatomy of the hippocampus using the ABA dataset. In addition to manual analysis, 1.89 two clustering methods were employed, a modified Non-negative Matrix Factorization (NNMF), and a hierarchial recursive 1.90 -bifurcation clustering scheme based on correlation as the similarity score. The paper yielded impressive results, proving the 1.91 -usefulness of such research. We have run NNMF on the cortical dataset5 and while the results are promising (see Preliminary 1.92 -Data), we think that it will be possible to find an even better method. In addition, this paper described a visual screening 1.93 -of the data, specifically, a visual analysis of 6000 genes with the primary purpose of observing how the spatial pattern of 1.94 -their expression coincided with the regions that had been identified by NNMF. We propose to do this sort of screening 1.95 -automatically, which would yield an objective, quantifiable result, rather than qualitative observations. 1.96 -AGEA’s[2] hierarchial clustering differs from our Aim 2 in at least two ways. First, AGEA uses perhaps the simplest 1.97 +bifurcation clustering scheme based on correlation as the similarity score. The paper yielded impressive results, proving 1.98 +the usefulness of computational genomic anatomy. We have run NNMF on the cortical dataset6 and while the results are 1.99 +promising (see Preliminary Data), we think that it will be possible to find an even better method. 1.100 +AGEA’s[6] hierarchial clustering differs from our Aim 2 in at least two ways. First, AGEA uses perhaps the simplest 1.101 possible similarity score (correlation), and does no dimensionality reduction before calculating similarity. While it is possible 1.102 that a more complex system will not do any better than this, we believe further exploration of alternative methods of scoring 1.103 and dimensionality reduction is warranted. Second, AGEA did not look at clusters of genes; in Preliminary Data we have 1.104 shown that clusters of genes may identify interesting spatial regions such as cortical areas. 1.105 -[? ] todo 1.106 +[10 ] todo 1.107 In summary, although these projects obtained hierarchial clusterings, there has not been much comparison between 1.108 different algorithms or scoring methods, so it is likely that the best clustering method for this application has not yet been 1.109 found. 1.110 @@ -211,8 +213,8 @@ 1.111 their approximate location upon the cortical surface. 1.112 Even the questions of how many areas should be recognized in cortex, and what their arrangement is, are still not 1.113 completely settled. A proposed division of the cortex into areas is called a cortical map. In the rodent, the lack of a 1.114 -single agreed-upon map can be seen by contrasting the recent maps given by Swanson[4] on the one hand, and Paxinos 1.115 -and Franklin[3] on the other. While the maps are certainly very similar in their general arrangement, significant differences 1.116 +single agreed-upon map can be seen by contrasting the recent maps given by Swanson[8] on the one hand, and Paxinos 1.117 +and Franklin[7] on the other. While the maps are certainly very similar in their general arrangement, significant differences 1.118 remain in the details. 1.119 The Allen Mouse Brain Atlas dataset 1.120 The Allen Mouse Brain Atlas (ABA) data was produced by doing in-situ hybridization on slices of male, 56-day-old 1.121 @@ -222,21 +224,21 @@ 1.122 mouse brains were needed in order to measure the expression of many genes. 1.123 Next, an automated nonlinear alignment procedure located the 2D data from the various slices in a single 3D coordinate 1.124 system. In the final 3D coordinate system, voxels are cubes with 200 microns on a side. There are 67x41x58 = 159,326 1.125 -voxels in the 3D coordinate system, of which 51,533 are in the brain[2]. 1.126 -Mus musculus, the common house mouse, is thought to contain about 22,000 protein-coding genes[6]. The ABA contains 1.127 +voxels in the 3D coordinate system, of which 51,533 are in the brain[6]. 1.128 +Mus musculus, the common house mouse, is thought to contain about 22,000 protein-coding genes[12]. The ABA contains 1.129 data on about 20,000 genes in sagittal sections, out of which over 4,000 genes are also measured in coronal sections. Our 1.130 dataset is derived from only the coronal subset of the ABA, because the sagittal data does not cover the entire cortex, 1.131 -and has greater registration error[2]. Genes were selected by the Allen Institute for coronal sectioning based on, “classes of 1.132 -known neuroscientific interest... or through post hoc identification of a marked non-ubiquitous expression pattern”[2]. 1.133 -The ABA is not the only large public spatial gene expression dataset. Other such resources include GENSAT[?], 1.134 -GenePaint[?], its sister project GeneAtlas[?], BGEM[?], EMAGE[?], EurExpress (http://www.eurexpress.org/ee/; Eur- 1.135 -Express data is also entered into EMAGE), todo. With the exception of the ABA, GenePaint, and EMAGE, most of these 1.136 +and has greater registration error[6]. Genes were selected by the Allen Institute for coronal sectioning based on, “classes of 1.137 +known neuroscientific interest... or through post hoc identification of a marked non-ubiquitous expression pattern”[6]. 1.138 +The ABA is not the only large public spatial gene expression dataset. Other such resources include GENSAT[3], 1.139 +GenePaint[11], its sister project GeneAtlas[1], BGEM[5], EMAGE[?], EurExpress (http://www.eurexpress.org/ee/; Eu- 1.140 +rExpress data is also entered into EMAGE), todo. With the exception of the ABA, GenePaint, and EMAGE, most of these 1.141 resources, have not (yet) extracted the expression intensity from the ISH images and registered the results into a single 3-D 1.142 space, and only ABA and EMAGE make this form of data available for public download from the website. Many of these 1.143 resources focus on developmental gene expression. 1.144 Significance 1.145 ___________________________ 1.146 - 5We ran “vanilla” NNMF, whereas the paper under discussion used a modified method. Their main modification consisted of adding a soft 1.147 + 6We ran “vanilla” NNMF, whereas the paper under discussion used a modified method. Their main modification consisted of adding a soft 1.148 spatial contiguity constraint. However, on our dataset, NNMF naturally produced spatially contiguous clusters, so no additional constraint was 1.149 needed. The paper under discussion also mentions that they tried a hierarchial variant of NNMF, which we have not yet tried. 1.150 The method developed in aim (1) will be applied to each cortical area to find a set of marker genes such that the 1.151 @@ -257,17 +259,12 @@ 1.152 While we do not here propose to analyze human gene expression data, it is conceivable that the methods we propose to 1.153 develop could be used to suggest modifications to the human cortical map as well. 1.154 Related work 1.155 -[2 ] describes the application of AGEA to the cortex. The paper describes interesting results on the structure of correlations 1.156 +[6 ] describes the application of AGEA to the cortex. The paper describes interesting results on the structure of correlations 1.157 between voxel gene expression profiles within a handful of cortical areas. However, this sort of analysis is not related to 1.158 either of our aims, as it neither finds marker genes, nor does it suggest a cortical map based on gene expression data. Neither 1.159 of the other components of AGEA can be applied to cortical areas; AGEA’s Gene Finder cannot be used to find marker 1.160 -genes for cortical areas; and AGEA’s hierarchial clustering does not produce clusters corresponding to cortical areas. 1.161 -In both cases, the root cause is that pairwise correlations between the gene expression of voxels in different areas but 1.162 -the same layer are stronger than pairwise correlations between the gene expression of voxels in different layers but the same 1.163 -area. Therefore a pairwise voxel correlation clustering algorithm will always create clusters representing cortical layers, not 1.164 -areas. This is why the hierarchial clustering does not find cortical areas6. The reason that Gene Finder cannot find marker 1.165 -genes for cortical areas is that in Gene Finder, although the user chooses a seed voxel, Gene Finder chooses the ROI for 1.166 -which genes will be found, and it creates that ROI by (pairwise voxel correlation) clustering around the seed. 1.167 +genes for most cortical areas; and AGEA’s hierarchial clustering does not produce clusters corresponding to most cortical 1.168 +areas7 . 1.169 In summary, for all three aims, (a) none of the previous projects explores combinations of marker genes, (b) there has 1.170 been almost no comparison of different algorithms or scoring methods, and (c) there has been no work on computationally 1.171 finding marker genes for cortical areas, or on finding a hierarchial clustering that will yield a map of cortical areas de novo 1.172 @@ -275,8 +272,13 @@ 1.173 Our project is guided by a concrete application with a well-specified criterion of success (how well we can find marker 1.174 genes for / reproduce the layout of cortical areas), which will provide a solid basis for comparing different methods. 1.175 _________________________________________ 1.176 - 6There are clusters which presumably correspond to the intersection of a layer and an area, but since one area will have many layer-area 1.177 -intersection clusters, further work is needed to make sense of these. 1.178 + 7In both cases, the root cause is that pairwise correlations between the gene expression of voxels in different areas but the same layer are 1.179 +often stronger than pairwise correlations between the gene expression of voxels in different layers but the same area. Therefore, a pairwise voxel 1.180 +correlation clustering algorithm will often create clusters representing cortical layers, not areas. This is why the hierarchial clustering does not 1.181 +find most cortical areas (there are clusters which presumably correspond to the intersection of a layer and an area, but since one area will have 1.182 +many layer-area intersection clusters, further work is needed to make sense of these). The reason that Gene Finder cannot find marker genes for 1.183 +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, 1.184 +and it creates that ROI by (pairwise voxel correlation) clustering around the seed. 1.185 Preliminary work 1.186 Format conversion between SEV, MATLAB, NIFTI 1.187 We have created software to (politely) download all of the SEV files from the Allen Institute website. We have also created 1.188 @@ -284,7 +286,7 @@ 1.189 Flatmap of cortex 1.190 We downloaded the ABA data and applied a mask to select only those voxels which belong to cerebral cortex. We divided 1.191 the cortex into hemispheres. 1.192 -Using Caret[1], we created a mesh representation of the surface of the selected voxels. For each gene, for each node of 1.193 +Using Caret[2], we created a mesh representation of the surface of the selected voxels. For each gene, for each node of 1.194 the mesh, we calculated an average of the gene expression of the voxels “underneath” that mesh node. We then flattened 1.195 the cortex, creating a two-dimensional mesh. 1.196 We sampled the nodes of the irregular, flat mesh in order to create a regular grid of pixel values. We converted this grid 1.197 @@ -357,8 +359,8 @@ 1.198 similar direction (because the borders are similar). 1.199 Gradient similarity provides information complementary to correlation 1.200 To show that gradient similarity can provide useful information that cannot be detected via pointwise analyses, consider 1.201 -Fig. . The top row of Fig. displays the 3 genes which most match area AUD, according to a pointwise method7. The bottom 1.202 -row displays the 3 genes which most match AUD according to a method which considers local geometry8 The pointwise 1.203 +Fig. . The top row of Fig. displays the 3 genes which most match area AUD, according to a pointwise method8. The bottom 1.204 +row displays the 3 genes which most match AUD according to a method which considers local geometry9 The pointwise 1.205 method in the top row identifies genes which express more strongly in AUD than outside of it; its weakness is that this 1.206 includes many areas which don’t have a salient border matching the areal border. The geometric method identifies genes 1.207 whose salient expression border seems to partially line up with the border of AUD; its weakness is that this includes genes 1.208 @@ -367,14 +369,14 @@ 1.209 we deliberately chose a “difficult” area in order to better contrast pointwise with geometric methods. 1.210 Combinations of multiple genes are useful 1.211 Here we give an example of a cortical area which is not marked by any single gene, but which can be identified combi- 1.212 -natorially. according to logistic regression, gene wwc19 is the best fit single gene for predicting whether or not a pixel on 1.213 +natorially. according to logistic regression, gene wwc110 is the best fit single gene for predicting whether or not a pixel on 1.214 _________________________________________ 1.215 - 7For each gene, a logistic regression in which the response variable was whether or not a surface pixel was within area AUD, and the predictor 1.216 + 8For each gene, a logistic regression in which the response variable was whether or not a surface pixel was within area AUD, and the predictor 1.217 variable was the value of the expression of the gene underneath that pixel. The resulting scores were used to rank the genes in terms of how well 1.218 they predict area AUD. 1.219 - 8For each gene the gradient similarity (see section ??) between (a) a map of the expression of each gene on the cortical surface and (b) the 1.220 + 9For each gene the gradient similarity (see section ??) between (a) a map of the expression of each gene on the cortical surface and (b) the 1.221 shape of area AUD, was calculated, and this was used to rank the genes. 1.222 - 9“WW, C2 and coiled-coil domain containing 1”; EntrezGene ID 211652 1.223 + 10“WW, C2 and coiled-coil domain containing 1”; EntrezGene ID 211652 1.224 1.225 1.226 1.227 @@ -387,7 +389,7 @@ 1.228 pattern over the cortex. The lower-right boundary of MO is represented reasonably well by this gene, however the gene 1.229 overshoots the upper-left boundary. This flattened 2-D representation does not show it, but the area corresponding to the 1.230 overshoot is the medial surface of the cortex. MO is only found on the lateral surface (todo). 1.231 -Gene mtif210 is shown in figure the upper-right of Fig. . Mtif2 captures MO’s upper-left boundary, but not its lower-right 1.232 +Gene mtif211 is shown in figure the upper-right of Fig. . Mtif2 captures MO’s upper-left boundary, but not its lower-right 1.233 boundary. Mtif2 does not express very much on the medial surface. By adding together the values at each pixel in these 1.234 two figures, we get the lower-left of Figure . This combination captures area MO much better than any single gene. 1.235 Areas which can be identified by single genes 1.236 @@ -398,7 +400,7 @@ 1.237 Forward stepwise logistic regression todo 1.238 SVM on all genes at once 1.239 In order to see how well one can do when looking at all genes at once, we ran a support vector machine to classify cortical 1.240 -surface pixels based on their gene expression profiles. We achieved classification accuracy of about 81%11. As noted above, 1.241 +surface pixels based on their gene expression profiles. We achieved classification accuracy of about 81%12. As noted above, 1.242 however, a classifier that looks at all the genes at once isn’t practically useful. 1.243 The requirement to find combinations of only a small number of genes limits us from straightforwardly applying many 1.244 of the most simple techniques from the field of supervised machine learning. In the parlance of machine learning, our task 1.245 @@ -410,8 +412,8 @@ 1.246 todo 1.247 (might want to incld nnMF since mentioned above) 1.248 _________________________________________ 1.249 - 10“mitochondrial translational initiation factor 2”; EntrezGene ID 76784 1.250 - 115-fold cross-validation. 1.251 + 11“mitochondrial translational initiation factor 2”; EntrezGene ID 76784 1.252 + 125-fold cross-validation. 1.253 Dimensionality reduction plus K-means or spectral clustering 1.254 Many areas are captured by clusters of genes 1.255 todo 1.256 @@ -422,7 +424,7 @@ 1.257 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 1.258 in the latter case it is curved. If the manifold is curved, there are various methods for mapping the manifold into a plane. 1.259 In the case of the cerebral cortex, it remains to be seen which method of mapping the manifold into a plane is optimal 1.260 -for this application. We will compare mappings which attempt to preserve size (such as the one used by Caret[1]) with 1.261 +for this application. We will compare mappings which attempt to preserve size (such as the one used by Caret[2]) with 1.262 mappings which preserve angle (conformal maps). 1.263 Although there is much 2-D organization in anatomy, there are also structures whose shape is fundamentally 3-dimensional. 1.264 If possible, we would like the method we develop to include a statistical test that warns the user if the assumption of 2-D 1.265 @@ -458,52 +460,70 @@ 1.266 5.Develop an algorithm to use dimensionality reduction and/or hierarchial clustering to create anatomical maps 1.267 6.Run this algorithm on the cortex: present a hierarchial, genoarchitectonic map of the cortex 1.268 Bibliography & References Cited 1.269 -[1]D C Van Essen, H A Drury, J Dickson, J Harwell, D Hanlon, and C H Anderson. An integrated software suite for surface- 1.270 +[1]J. Carson, T. Ju, C. Thaller, M. Bello, I. Kakadiaris, J. Warren, G. Eichele, and W. Chiu. Data mining in situ gene 1.271 +expression patterns at cellular resolution. In Computational Systems Bioinformatics Conference, 2005. Workshops and 1.272 +Poster Abstracts. IEEE, page 358, 2005. 1.273 +[2]D C Van Essen, H A Drury, J Dickson, J Harwell, D Hanlon, and C H Anderson. An integrated software suite for surface- 1.274 based analyses of cerebral cortex. Journal of the American Medical Informatics Association: JAMIA, 8(5):443–59, 2001. 1.275 PMID: 11522765. 1.276 -[2]Lydia Ng, Amy Bernard, Chris Lau, Caroline C Overly, Hong-Wei Dong, Chihchau Kuan, Sayan Pathak, Susan M Sunkin, 1.277 -Chinh Dang, Jason W Bohland, Hemant Bokil, Partha P Mitra, Luis Puelles, John Hohmann, David J Anderson, Ed S 1.278 -Lein, Allan R Jones, and Michael Hawrylycz. An anatomic gene expression atlas of the adult mouse brain. Nat Neurosci, 1.279 -12(3):356–362, March 2009. 1.280 -[3]George Paxinos and Keith B.J. Franklin. The Mouse Brain in Stereotaxic Coordinates. Academic Press, 2 edition, July 1.281 +[3]Shiaoching Gong, Chen Zheng, Martin L. Doughty, Kasia Losos, Nicholas Didkovsky, Uta B. Schambra, Norma J. 1.282 +Nowak, Alexandra Joyner, Gabrielle Leblanc, Mary E. Hatten, and Nathaniel Heintz. A gene expression atlas of the 1.283 +central nervous system based on bacterial artificial chromosomes. Nature, 425(6961):917–925, October 2003. 1.284 +[4]Jano Hemert and Richard Baldock. Matching Spatial Regions with Combinations of Interacting Gene Expression 1.285 +Patterns, pages 347–361. 2008. 1.286 +[5]Susan Magdaleno, Patricia Jensen, Craig L. Brumwell, Anna Seal, Karen Lehman, Andrew Asbury, Tony Cheung, 1.287 +Tommie Cornelius, Diana M. Batten, Christopher Eden, Shannon M. Norland, Dennis S. Rice, Nilesh Dosooye, Sundeep 1.288 +Shakya, Perdeep Mehta, and Tom Curran. BGEM: an in situ hybridization database of gene expression in the embryonic 1.289 +and adult mouse nervous system. PLoS Biology, 4(4):e86 EP –, April 2006. 1.290 +[6]Lydia Ng, Amy Bernard, Chris Lau, Caroline C Overly, Hong-Wei Dong, Chihchau Kuan, Sayan Pathak, Susan M 1.291 +Sunkin, Chinh Dang, Jason W Bohland, Hemant Bokil, Partha P Mitra, Luis Puelles, John Hohmann, David J Anderson, 1.292 +Ed S Lein, Allan R Jones, and Michael Hawrylycz. An anatomic gene expression atlas of the adult mouse brain. Nat 1.293 +Neurosci, 12(3):356–362, March 2009. 1.294 +[7]George Paxinos and Keith B.J. Franklin. The Mouse Brain in Stereotaxic Coordinates. Academic Press, 2 edition, July 1.295 2001. 1.296 -[4]Larry Swanson. Brain Maps: Structure of the Rat Brain. Academic Press, 3 edition, November 2003. 1.297 -[5]Carol L. Thompson, Sayan D. Pathak, Andreas Jeromin, Lydia L. Ng, Cameron R. MacPherson, Marty T. Mortrud, 1.298 +[8]Larry Swanson. Brain Maps: Structure of the Rat Brain. Academic Press, 3 edition, November 2003. 1.299 +[9]Carol L. Thompson, Sayan D. Pathak, Andreas Jeromin, Lydia L. Ng, Cameron R. MacPherson, Marty T. Mortrud, 1.300 Allison Cusick, Zackery L. Riley, Susan M. Sunkin, Amy Bernard, Ralph B. Puchalski, Fred H. Gage, Allan R. Jones, 1.301 Vladimir B. Bajic, Michael J. Hawrylycz, and Ed S. Lein. Genomic anatomy of the hippocampus. Neuron, 60(6):1010– 1.302 1021, December 2008. 1.303 -[6]Robert H Waterston, Kerstin Lindblad-Toh, Ewan Birney, Jane Rogers, Josep F Abril, Pankaj Agarwal, Richa Agarwala, 1.304 -Rachel Ainscough, Marina Alexandersson, Peter An, Stylianos E Antonarakis, John Attwood, Robert Baertsch, Jonathon 1.305 -Bailey, Karen Barlow, Stephan Beck, Eric Berry, Bruce Birren, Toby Bloom, Peer Bork, Marc Botcherby, Nicolas Bray, 1.306 -Michael R Brent, Daniel G Brown, Stephen D Brown, Carol Bult, John Burton, Jonathan Butler, Robert D Campbell, 1.307 -Piero Carninci, Simon Cawley, Francesca Chiaromonte, Asif T Chinwalla, Deanna M Church, Michele Clamp, Christopher 1.308 -Clee, Francis S Collins, Lisa L Cook, Richard R Copley, Alan Coulson, Olivier Couronne, James Cuff, Val Curwen, Tim 1.309 -Cutts, Mark Daly, Robert David, Joy Davies, Kimberly D Delehaunty, Justin Deri, Emmanouil T Dermitzakis, Colin 1.310 -Dewey, Nicholas J Dickens, Mark Diekhans, Sheila Dodge, Inna Dubchak, Diane M Dunn, Sean R Eddy, Laura Elnitski, 1.311 -Richard D Emes, Pallavi Eswara, Eduardo Eyras, Adam Felsenfeld, Ginger A Fewell, Paul Flicek, Karen Foley, Wayne N 1.312 -Frankel, Lucinda A Fulton, Robert S Fulton, Terrence S Furey, Diane Gage, Richard A Gibbs, Gustavo Glusman, Sante 1.313 -Gnerre, Nick Goldman, Leo Goodstadt, Darren Grafham, Tina A Graves, Eric D Green, Simon Gregory, Roderic Guig, 1.314 -Mark Guyer, Ross C Hardison, David Haussler, Yoshihide Hayashizaki, LaDeana W Hillier, Angela Hinrichs, Wratko 1.315 -Hlavina, Timothy Holzer, Fan Hsu, Axin Hua, Tim Hubbard, Adrienne Hunt, Ian Jackson, David B Jaffe, L Steven 1.316 -Johnson, Matthew Jones, Thomas A Jones, Ann Joy, Michael Kamal, Elinor K Karlsson, Donna Karolchik, Arkadiusz 1.317 -Kasprzyk, Jun Kawai, Evan Keibler, Cristyn Kells, W James Kent, Andrew Kirby, Diana L Kolbe, Ian Korf, Raju S 1.318 -Kucherlapati, Edward J Kulbokas, David Kulp, Tom Landers, J P Leger, Steven Leonard, Ivica Letunic, Rosie Levine, Jia 1.319 -Li, Ming Li, Christine Lloyd, Susan Lucas, Bin Ma, Donna R Maglott, Elaine R Mardis, Lucy Matthews, Evan Mauceli, 1.320 -John H Mayer, Megan McCarthy, W Richard McCombie, Stuart McLaren, Kirsten McLay, John D McPherson, Jim 1.321 -Meldrim, Beverley Meredith, Jill P Mesirov, Webb Miller, Tracie L Miner, Emmanuel Mongin, Kate T Montgomery, 1.322 -Michael Morgan, Richard Mott, James C Mullikin, Donna M Muzny, William E Nash, Joanne O Nelson, Michael N 1.323 -Nhan, Robert Nicol, Zemin Ning, Chad Nusbaum, Michael J O’Connor, Yasushi Okazaki, Karen Oliver, Emma Overton- 1.324 -Larty, Lior Pachter, Gens Parra, Kymberlie H Pepin, Jane Peterson, Pavel Pevzner, Robert Plumb, Craig S Pohl, Alex 1.325 -Poliakov, Tracy C Ponce, Chris P Ponting, Simon Potter, Michael Quail, Alexandre Reymond, Bruce A Roe, Krishna M 1.326 -Roskin, Edward M Rubin, Alistair G Rust, Ralph Santos, Victor Sapojnikov, Brian Schultz, Jrg Schultz, Matthias S 1.327 -Schwartz, Scott Schwartz, Carol Scott, Steven Seaman, Steve Searle, Ted Sharpe, Andrew Sheridan, Ratna Shownkeen, 1.328 -Sarah Sims, Jonathan B Singer, Guy Slater, Arian Smit, Douglas R Smith, Brian Spencer, Arne Stabenau, Nicole Stange- 1.329 -Thomann, Charles Sugnet, Mikita Suyama, Glenn Tesler, Johanna Thompson, David Torrents, Evanne Trevaskis, John 1.330 -Tromp, Catherine Ucla, Abel Ureta-Vidal, Jade P Vinson, Andrew C Von Niederhausern, Claire M Wade, Melanie Wall, 1.331 -Ryan J Weber, Robert B Weiss, Michael C Wendl, Anthony P West, Kris Wetterstrand, Raymond Wheeler, Simon 1.332 -Whelan, Jamey Wierzbowski, David Willey, Sophie Williams, Richard K Wilson, Eitan Winter, Kim C Worley, Dudley 1.333 -Wyman, Shan Yang, Shiaw-Pyng Yang, Evgeny M Zdobnov, Michael C Zody, and Eric S Lander. Initial sequencing and 1.334 -comparative analysis of the mouse genome. Nature, 420(6915):520–62, December 2002. PMID: 12466850. 1.335 +[10]Shanmugasundaram Venkataraman, Peter Stevenson, Yiya Yang, Lorna Richardson, Nicholas Burton, Thomas P. Perry, 1.336 +Paul Smith, Richard A. Baldock, Duncan R. Davidson, and Jeffrey H. Christiansen. EMAGE edinburgh mouse atlas 1.337 +of gene expression: 2008 update. Nucl. Acids Res., 36(suppl_1):D860–865, 2008. 1.338 +[11]Axel Visel, Christina Thaller, and Gregor Eichele. GenePaint.org: an atlas of gene expression patterns in the mouse 1.339 +embryo. Nucl. Acids Res., 32(suppl_1):D552–556, 2004. 1.340 +[12]Robert H Waterston, Kerstin Lindblad-Toh, Ewan Birney, Jane Rogers, Josep F Abril, Pankaj Agarwal, Richa Agar- 1.341 +wala, Rachel Ainscough, Marina Alexandersson, Peter An, Stylianos E Antonarakis, John Attwood, Robert Baertsch, 1.342 +Jonathon Bailey, Karen Barlow, Stephan Beck, Eric Berry, Bruce Birren, Toby Bloom, Peer Bork, Marc Botcherby, 1.343 +Nicolas Bray, Michael R Brent, Daniel G Brown, Stephen D Brown, Carol Bult, John Burton, Jonathan Butler, 1.344 +Robert D Campbell, Piero Carninci, Simon Cawley, Francesca Chiaromonte, Asif T Chinwalla, Deanna M Church, 1.345 +Michele Clamp, Christopher Clee, Francis S Collins, Lisa L Cook, Richard R Copley, Alan Coulson, Olivier Couronne, 1.346 +James Cuff, Val Curwen, Tim Cutts, Mark Daly, Robert David, Joy Davies, Kimberly D Delehaunty, Justin Deri, 1.347 +Emmanouil T Dermitzakis, Colin Dewey, Nicholas J Dickens, Mark Diekhans, Sheila Dodge, Inna Dubchak, Diane M 1.348 +Dunn, Sean R Eddy, Laura Elnitski, Richard D Emes, Pallavi Eswara, Eduardo Eyras, Adam Felsenfeld, Ginger A 1.349 +Fewell, Paul Flicek, Karen Foley, Wayne N Frankel, Lucinda A Fulton, Robert S Fulton, Terrence S Furey, Diane Gage, 1.350 +Richard A Gibbs, Gustavo Glusman, Sante Gnerre, Nick Goldman, Leo Goodstadt, Darren Grafham, Tina A Graves, 1.351 +Eric D Green, Simon Gregory, Roderic Guig, Mark Guyer, Ross C Hardison, David Haussler, Yoshihide Hayashizaki, 1.352 +LaDeana W Hillier, Angela Hinrichs, Wratko Hlavina, Timothy Holzer, Fan Hsu, Axin Hua, Tim Hubbard, Adrienne 1.353 +Hunt, Ian Jackson, David B Jaffe, L Steven Johnson, Matthew Jones, Thomas A Jones, Ann Joy, Michael Kamal, 1.354 +Elinor K Karlsson, Donna Karolchik, Arkadiusz Kasprzyk, Jun Kawai, Evan Keibler, Cristyn Kells, W James Kent, 1.355 +Andrew Kirby, Diana L Kolbe, Ian Korf, Raju S Kucherlapati, Edward J Kulbokas, David Kulp, Tom Landers, J P 1.356 +Leger, Steven Leonard, Ivica Letunic, Rosie Levine, Jia Li, Ming Li, Christine Lloyd, Susan Lucas, Bin Ma, Donna R 1.357 +Maglott, Elaine R Mardis, Lucy Matthews, Evan Mauceli, John H Mayer, Megan McCarthy, W Richard McCombie, 1.358 +Stuart McLaren, Kirsten McLay, John D McPherson, Jim Meldrim, Beverley Meredith, Jill P Mesirov, Webb Miller, 1.359 +Tracie L Miner, Emmanuel Mongin, Kate T Montgomery, Michael Morgan, Richard Mott, James C Mullikin, Donna M 1.360 +Muzny, William E Nash, Joanne O Nelson, Michael N Nhan, Robert Nicol, Zemin Ning, Chad Nusbaum, Michael J 1.361 +O’Connor, Yasushi Okazaki, Karen Oliver, Emma Overton-Larty, Lior Pachter, Gens Parra, Kymberlie H Pepin, Jane 1.362 +Peterson, Pavel Pevzner, Robert Plumb, Craig S Pohl, Alex Poliakov, Tracy C Ponce, Chris P Ponting, Simon Potter, 1.363 +Michael Quail, Alexandre Reymond, Bruce A Roe, Krishna M Roskin, Edward M Rubin, Alistair G Rust, Ralph San- 1.364 +tos, Victor Sapojnikov, Brian Schultz, Jrg Schultz, Matthias S Schwartz, Scott Schwartz, Carol Scott, Steven Seaman, 1.365 +Steve Searle, Ted Sharpe, Andrew Sheridan, Ratna Shownkeen, Sarah Sims, Jonathan B Singer, Guy Slater, Arian 1.366 +Smit, Douglas R Smith, Brian Spencer, Arne Stabenau, Nicole Stange-Thomann, Charles Sugnet, Mikita Suyama, 1.367 +Glenn Tesler, Johanna Thompson, David Torrents, Evanne Trevaskis, John Tromp, Catherine Ucla, Abel Ureta-Vidal, 1.368 +Jade P Vinson, Andrew C Von Niederhausern, Claire M Wade, Melanie Wall, Ryan J Weber, Robert B Weiss, Michael C 1.369 +Wendl, Anthony P West, Kris Wetterstrand, Raymond Wheeler, Simon Whelan, Jamey Wierzbowski, David Willey, 1.370 +Sophie Williams, Richard K Wilson, Eitan Winter, Kim C Worley, Dudley Wyman, Shan Yang, Shiaw-Pyng Yang, 1.371 +Evgeny M Zdobnov, Michael C Zody, and Eric S Lander. Initial sequencing and comparative analysis of the mouse 1.372 +genome. Nature, 420(6915):520–62, December 2002. PMID: 12466850. 1.373 1.374 _______________________________________________________________________________________________________ 1.375 stuff i dunno where to put yet (there is more scattered through grant-oldtext): 1.376 @@ -512,6 +532,5 @@ 1.377 note: 1.378 do we need to cite: no known markers, impressive results? 1.379 two hemis 1.380 - “genomic anatomy” is a name found in the titles of one of the cited papers which seems good 1.381 1.382
2.1 Binary file grant.odt has changed
3.1 Binary file grant.pdf has changed
4.1 --- a/grant.txt Wed Apr 15 00:50:34 2009 -0700 4.2 +++ b/grant.txt Wed Apr 15 03:19:01 2009 -0700 4.3 @@ -67,11 +67,11 @@ 4.4 4.5 4.6 === Related work === 4.7 -There is a substantial body of work on the analysis of gene expression data, however, most of this concerns gene expression data which is not fundamentally spatial. 4.8 +There is a substantial body of work on the analysis of gene expression data, most of this concerns gene expression data which is not fundamentally spatial\footnote{By "__fundamentally__ spatial" we mean that there is information from a large number of spatial locations; not just data which has only a few different locations.}. 4.9 4.10 As noted above, there has been much work on both supervised learning and there are many available algorithms for each. However, the algorithms require the scientist to provide a framework for representing the problem domain, and the way that this framework is set up has a large impact on performance. Creating a good framework can require creatively reconceptualizing the problem domain, and is not merely a mechanical "fine-tuning" of numerical parameters. For example, we believe that domain-specific scoring measures (such as gradient similarity, which is discussed in Preliminary Work) may be necessary in order to achieve the best results in this application. 4.11 4.12 -We are aware of three existing efforts to find marker genes using spatial gene expression data using automated methods. 4.13 +We are aware of four existing efforts to find marker genes using spatial gene expression data using automated methods. 4.14 4.15 \cite{carson_data_2005} describes GeneAtlas. GeneAtlas allows the user to construct a search query by freely demarcating one or two 2-D regions on sagittal slices, and then to specify either the strength of expression or the name of another gene whose expression pattern is to be matched. GeneAtlas differs from our Aim 1 in at least two ways. First, GeneAtlas finds only single genes, whereas we will also look for combinations of genes\footnote{See Preliminary Data for an example of an area which cannot be marked by any single gene in the dataset, but which can be marked by a combination.}. Second, at least for the custom spatial search, Gene Atlas appears to use a simple pointwise scoring method (strength of expression), whereas we will also use geometric metrics such as gradient similarity. 4.16 4.17 @@ -91,6 +91,11 @@ 4.18 4.19 Gene Finder is different from our Aim 1 in at least three ways. First, Gene Finder finds only single genes, whereas we will also look for combinations of genes. Second, gene finder can only use overexpression as a marker, whereas we will also search for underexpression. Third, Gene Finder uses a simple pointwise score\footnote{"Expression energy ratio", which captures overexpression.}, whereas we will also use geometric scores such as gradient similarity. The Preliminary Data section contains evidence that each of our three choices is the right one. 4.20 4.21 +\cite{venkataraman_emage_2008} todo 4.22 + 4.23 + 4.24 +\cite{hemert_matching_2008} todo 4.25 + 4.26 In summary, none of the previous projects explores combinations of marker genes, and none of their publications compare the results obtained by using different algorithms or scoring methods. 4.27 4.28 4.29 @@ -135,7 +140,7 @@ 4.30 4.31 Gene clusters could be used as part of dimensionality reduction: rather than have one feature for each gene, we could have one reduced feature for each gene cluster. 4.32 4.33 -Gene clusters could also be used to directly yield a clustering on instances. This is because many genes have an expression pattern which seems to pick out a single, spatially continguous region. Therefore, it seems likely that an anatomically interesting region will have multiple genes which each individually pick it out\footnote{This would seem to contradict our finding in aim 1 that some cortical areas are combinatorially coded by multiple genes. However, it is possible that the currently accepted cortical maps divide the cortex into regions which are unnatural from the point of view of gene expression; perhaps there is some other way to map the cortex for which each region can be identified by single genes.}. This suggests the following procedure: cluster together genes which pick out similar regions, and then to use the more popular common regions as the final clusters. In the Preliminary Data we show that a number of anatomically recognized cortical regions, as well as some "superregions" formed by lumping together a few regions, are associated with gene clusters in this fashion. 4.34 +Gene clusters could also be used to directly yield a clustering on instances. This is because many genes have an expression pattern which seems to pick out a single, spatially continguous region. Therefore, it seems likely that an anatomically interesting region will have multiple genes which each individually pick it out\footnote{This would seem to contradict our finding in aim 1 that some cortical areas are combinatorially coded by multiple genes. However, it is possible that the currently accepted cortical maps divide the cortex into regions which are unnatural from the point of view of gene expression; perhaps there is some other way to map the cortex for which each region can be identified by single genes. Another possibility is that, although the cluster prototype fits an anatomical region, the individual genes are each somewhat different from the prototype.}. This suggests the following procedure: cluster together genes which pick out similar regions, and then to use the more popular common regions as the final clusters. In the Preliminary Data we show that a number of anatomically recognized cortical regions, as well as some "superregions" formed by lumping together a few regions, are associated with gene clusters in this fashion. 4.35 4.36 4.37 === Related work === 4.38 @@ -145,7 +150,9 @@ 4.39 \cite{thompson_genomic_2008} describes an analysis of the anatomy of 4.40 the hippocampus using the ABA dataset. In addition to manual analysis, 4.41 two clustering methods were employed, a modified Non-negative Matrix 4.42 -Factorization (NNMF), and a hierarchial recursive bifurcation clustering scheme based on correlation as the similarity score. The paper yielded impressive results, proving the usefulness of such research. We have run NNMF on the cortical dataset\footnote{We ran "vanilla" NNMF, whereas the paper under discussion used a modified method. Their main modification consisted of adding a soft spatial contiguity constraint. However, on our dataset, NNMF naturally produced spatially contiguous clusters, so no additional constraint was needed. The paper under discussion also mentions that they tried a hierarchial variant of NNMF, which we have not yet tried.} and while the results are promising (see Preliminary Data), we think that it will be possible to find an even better method. In addition, this paper described a visual screening of the data, specifically, a visual analysis of 6000 genes with the primary purpose of observing how the spatial pattern of their expression coincided with the regions that had been identified by NNMF. We propose to do this sort of screening automatically, which would yield an objective, quantifiable result, rather than qualitative observations. 4.43 +Factorization (NNMF), and a hierarchial recursive bifurcation clustering scheme based on correlation as the similarity score. The paper yielded impressive results, proving the usefulness of computational genomic anatomy. We have run NNMF on the cortical dataset\footnote{We ran "vanilla" NNMF, whereas the paper under discussion used a modified method. Their main modification consisted of adding a soft spatial contiguity constraint. However, on our dataset, NNMF naturally produced spatially contiguous clusters, so no additional constraint was needed. The paper under discussion also mentions that they tried a hierarchial variant of NNMF, which we have not yet tried.} and while the results are promising (see Preliminary Data), we think that it will be possible to find an even better method. 4.44 + 4.45 +%% In addition, this paper described a visual screening of the data, specifically, a visual analysis of 6000 genes with the primary purpose of observing how the spatial pattern of their expression coincided with the regions that had been identified by NNMF. We propose to do this sort of screening automatically, which would yield an objective, quantifiable result, rather than qualitative observations. 4.46 4.47 4.48 4.49 @@ -181,7 +188,7 @@ 4.50 4.51 Mus musculus, the common house mouse, is thought to contain about 22,000 protein-coding genes\cite{waterston_initial_2002}. The ABA contains data on about 20,000 genes in sagittal sections, out of which over 4,000 genes are also measured in coronal sections. Our dataset is derived from only the coronal subset of the ABA, because the sagittal data does not cover the entire cortex, and has greater registration error\cite{ng_anatomic_2009}. Genes were selected by the Allen Institute for coronal sectioning based on, "classes of known neuroscientific interest... or through post hoc identification of a marked non-ubiquitous expression pattern"\cite{ng_anatomic_2009}. 4.52 4.53 -The ABA is not the only large public spatial gene expression dataset. Other such resources include GENSAT\cite{gong_gene_2003}, GenePaint\cite{visel_genepaint_2004}, its sister project GeneAtlas\cite{carson_data_2005}, BGEM\cite{magdaleno_bgem_2006}, EMAGE\cite{?}, EurExpress (http://www.eurexpress.org/ee/; EurExpress data is also entered into EMAGE), todo. With the exception of the ABA, GenePaint, and EMAGE, most of these resources, have not (yet) extracted the expression intensity from the ISH images and registered the results into a single 3-D space, and only ABA and EMAGE make this form of data available for public download from the website. Many of these resources focus on developmental gene expression. 4.54 +The ABA is not the only large public spatial gene expression dataset. Other such resources include GENSAT\cite{gong_gene_2003}, GenePaint\cite{visel_genepaint.org:atlas_2004}, its sister project GeneAtlas\cite{carson_data_2005}, BGEM\cite{magdaleno_bgem:in_2006}, EMAGE\cite{?}, EurExpress (http://www.eurexpress.org/ee/; EurExpress data is also entered into EMAGE), todo. With the exception of the ABA, GenePaint, and EMAGE, most of these resources, have not (yet) extracted the expression intensity from the ISH images and registered the results into a single 3-D space, and only ABA and EMAGE make this form of data available for public download from the website. Many of these resources focus on developmental gene expression. 4.55 4.56 4.57 4.58 @@ -198,15 +205,16 @@ 4.59 4.60 === Related work === 4.61 4.62 -\cite{ng_anatomic_2009} describes the application of AGEA to the cortex. The paper describes interesting results on the structure of correlations between voxel gene expression profiles within a handful of cortical areas. However, this sort of analysis is not related to either of our aims, as it neither finds marker genes, nor does it suggest a cortical map based on gene expression data. Neither of the other components of AGEA can be applied to cortical areas; AGEA's Gene Finder cannot be used to find marker genes for cortical areas; and AGEA's hierarchial clustering does not produce clusters corresponding to cortical areas. 4.63 - 4.64 -In both cases, the root cause is that pairwise correlations between the gene expression of voxels in different areas but the same layer are stronger than pairwise correlations between the gene expression of voxels in different layers but the same area. Therefore a pairwise voxel correlation clustering algorithm will always create clusters representing cortical layers, not areas. This is why the hierarchial clustering does not find cortical areas\footnote{There are clusters which presumably correspond to the intersection of a layer and an area, but since one area will have many layer-area intersection clusters, further work is needed to make sense of these.}. The reason that Gene Finder cannot find marker genes for 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, and it creates that ROI by (pairwise voxel correlation) clustering around the seed. 4.65 +\cite{ng_anatomic_2009} describes the application of AGEA to the cortex. The paper describes interesting results on the structure of correlations between voxel gene expression profiles within a handful of cortical areas. However, this sort of analysis is not related to either of our aims, as it neither finds marker genes, nor does it suggest a cortical map based on gene expression data. Neither of the other components of AGEA can be applied to cortical areas; AGEA's Gene Finder cannot be used to find marker genes for most cortical areas; and AGEA's hierarchial clustering does not produce clusters corresponding to most cortical areas\footnote{In both cases, the root cause is that pairwise correlations between the gene expression of voxels in different areas but the same layer are often stronger than pairwise correlations between the gene expression of voxels in different layers but the same area. Therefore, a pairwise voxel correlation clustering algorithm will often create clusters representing cortical layers, not areas. This is why the hierarchial clustering does not find most cortical areas (there are clusters which presumably correspond to the intersection of a layer and an area, but since one area will have many layer-area intersection clusters, further work is needed to make sense of these). The reason that Gene Finder cannot find marker genes for 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, and it creates that ROI by (pairwise voxel correlation) clustering around the seed.}. 4.66 + 4.67 4.68 In summary, for all three aims, (a) none of the previous projects explores combinations of marker genes, (b) there has been almost no comparison of different algorithms or scoring methods, and (c) there has been no work on computationally finding marker genes for cortical areas, or on finding a hierarchial clustering that will yield a map of cortical areas de novo from gene expression data. 4.69 4.70 Our project is guided by a concrete application with a well-specified criterion of success (how well we can find marker genes for \begin{latex}/\end{latex} reproduce the layout of cortical areas), which will provide a solid basis for comparing different methods. 4.71 4.72 4.73 +%% todo: poster; check AGEA cortical data 4.74 + 4.75 \newpage 4.76 4.77 == Preliminary work == 4.78 @@ -446,5 +454,6 @@ 4.79 two hemis 4.80 4.81 4.82 -"genomic anatomy" is a name found in the titles of one of the cited papers which seems good 4.83 - 4.84 +%%"genomic anatomy" is a name found in the titles of one of the cited papers which seems good; maybe "computational genomic anatomy" 4.85 + 4.86 +%% todo: actually i'm pretty sure AGEA doesn't find ANY areas, but i said "most" and "often" to be cautious.