PMC:2570073 / 830-7613 JSONTXT

{"project":"2_test","denotations":[{"id":"18982112-15043218-38187799","span":{"begin":231,"end":235},"obj":"15043218"},{"id":"18982112-11967564-38187800","span":{"begin":246,"end":250},"obj":"11967564"},{"id":"18982112-15043223-38187801","span":{"begin":298,"end":302},"obj":"15043223"},{"id":"18982112-15114360-38187802","span":{"begin":320,"end":324},"obj":"15114360"},{"id":"18982112-12374427-38187803","span":{"begin":332,"end":336},"obj":"12374427"},{"id":"18982112-16728446-38187804","span":{"begin":1920,"end":1924},"obj":"16728446"},{"id":"18982112-17521453-38187805","span":{"begin":1945,"end":1949},"obj":"17521453"},{"id":"18982112-16499990-38187806","span":{"begin":1959,"end":1963},"obj":"16499990"},{"id":"18982112-17911164-38187807","span":{"begin":1985,"end":1989},"obj":"17911164"},{"id":"18982112-17911164-38187808","span":{"begin":2787,"end":2791},"obj":"17911164"},{"id":"18982112-15043218-38187809","span":{"begin":3030,"end":3034},"obj":"15043218"},{"id":"18982112-17151600-38187810","span":{"begin":3049,"end":3053},"obj":"17151600"},{"id":"18982112-16184189-38187811","span":{"begin":3239,"end":3243},"obj":"16184189"},{"id":"18982112-16381949-38187812","span":{"begin":3266,"end":3270},"obj":"16381949"}],"text":"Atlases for Sharing Gene Expression Data\nMany neuroscientists have been calling for a system for the mouse brain where digital atlases serve as the framework used to traverse the brain and information linked to it (Baldock et al., 2003; Bjaalie, 2002; Boline et al., 2007; MacKenzie-Graham et al., 2003; Martone et al., 2004; Toga, 2002). In contrast to data repositories, which allow simple access to data through a single interface, sophisticated digital atlases backed by the appropriate technology can act as a neuroinformatics hub facilitating access to different databases, information sources, and related documents and annotations. They may act as the scaffold in which otherwise unrelated data may be housed and correlated, providing an intuitive interface to share, visualize, analyze, and mine data of multiple modalities, scales, and dimensions.\nThe semantic and spatial information tied to an atlas can add a dimension to data in a manner that exponentially increases its potential use and reusability. Semantic linking of data to the atlas requires the data provider to register it with an ontology or controlled vocabulary, while spatial registration requires alignment of an image to the atlas. This information is then used to place data into the context of the atlas, allowing it to inherit information tied to the spatial coordinates of the atlas. A perfect example of this is gene expression image data. Spatial information in this type of slice data is key to interpreting results, yet these images lose anatomical context during the data collection process. Thus atlas-based tools for organizing and analyzing this and related types of data may be used to create a system ideal for sharing data.\nSeveral projects offer access to gene expression image data with differing levels of spatial mapping in the mouse nervous system. These projects have been comprehensively reviewed (Brumwell and Curran, 2006; Koester and Insel, 2007; Sunkin, 2006; Sunkin and Hohmann, 2007) and among others, include the Allen Brain Atlas (ABA, www.brain-map.org), BGEM (www.stjudebgem.org), GENSAT (www.gensat.org), GenePaint (www.genepaint.org), EurExpress (www.eurexpress.org), MGI (http://www.informatics.jax.org/) and EMAP/EMAGE (http://genex.hgu.mrc.ac.uk). Several of these projects are geared toward the developmental stages of the mouse. As illustrated by the images from GENSAT and MGI in Figure 1, the ability to examine spatial and temporal expression patterns is crucial for developing correlations between genotype and phenotype as well as for interpreting and comparing findings across experiments. Also illustrated, is that the anatomical information tends to be sparse in these data sources, as is also the case with BGEM, GenePaint, and EurExpress (Sunkin and Hohmann, 2007). ABA and EMAP/EMAGE differ from these sources in that in addition to linking the images to semantic information, they have also linked their images to spatial information by registering their images to a reference atlas (Baldock et al., 2003; Lein et al., 2007). While this extra step can be both difficult and time consuming, it adds the potential for a great deal of analytical power and the ability to generate spatial queries (Carson et al., 2005; Christiansen et al., 2006; Leergaard and Bjaalie, 2007).\nFigure 1 Examples of databases which manage gene expression image without an anatomical framework. GENSAT and MGI both provide a rich repository of image data for gene expression. Neither uses a standard framework for data organization, rather, both describe the pattern and location of gene expression revealed by the image. GENSAT (A) describes the expression pattern for major structures in each image, while the MGI (B) summarizes a batch of assays (the MGI dataset shown here were used to derive the Lef1 gene examples shown in Lee et al., 2007). While these annotations aid interpretation, the ability to query or analyze these data in this format is very limited. As most resources are built for sharing a specific set of data, most discussed above are not yet set up to easily link their data to that offered by other groups, adding a barrier to analyzing data across experiments. Also, with the exception of EMAGE, mechanisms are not readily available for an individual to easily put data into a semantic and spatial framework that facilitates comparison of their own data to others. For these reasons, it still requires a great deal of research and work to compare data collected in different experiments, which is one of our communities greatest desires, but also most difficult tasks. These are some of the recent drivers for the call to create interoperability across data resources and offering access to this system to any scientist, within the context of a digital atlasing framework.\nIntegrating these and other data sources via such a framework would allow a researcher to easily query across these resources. One could look for studies of different collection modalities, strains, developmental stages, or disease models, or examine the expression patterns of genes or regions of interest across multiple studies. For example, a scientist investigating a disease model of Parkinson's with the microarray technique finds that an unexpected gene in the caudate/putamen area seems to have a reverse correlation with motor deficits. Wanting to know more, he uses this system to find data from other experiments that have examined the same disease model. He finds a MRI dataset illustrating a change in the shape and decrease in the volume of that area in later stages of the disease, and a high-resolution confocal dataset from this same region shows abnormal cell morphology with disease progression. In addition, he finds that expression of this gene in this region in normal animals decreases with age and that it is expressed at a higher level throughout life in a different mouse strain which shows resistance to Parkinson's. Compiling and analyzing data from these different experiments, many of which he did not even realize were applicable to his situation, allows him to more fully examine the potential role of this gene and to better inform his next experiments.\nWhile individual researchers often do a similar type of information gathering on their own, it can be difficult to examine other datasets, or we miss a relevant dataset because the data producers published it in relation to a very different topic. As diverse data generation continues to grow at an accelerated rate, we are in dire need of systems that make it easy for our community to contribute, organize, and find relevant data. While we are a long way from a fully implemented system that could perform the previous example, different groups have already created many of its components."}

{"project":"0_colil","denotations":[{"id":"18982112-15043218-243695","span":{"begin":231,"end":235},"obj":"15043218"},{"id":"18982112-11967564-243696","span":{"begin":246,"end":250},"obj":"11967564"},{"id":"18982112-15043223-243697","span":{"begin":298,"end":302},"obj":"15043223"},{"id":"18982112-15114360-243698","span":{"begin":320,"end":324},"obj":"15114360"},{"id":"18982112-12374427-243699","span":{"begin":332,"end":336},"obj":"12374427"},{"id":"18982112-16728446-243700","span":{"begin":1920,"end":1924},"obj":"16728446"},{"id":"18982112-17521453-243701","span":{"begin":1945,"end":1949},"obj":"17521453"},{"id":"18982112-16499990-243702","span":{"begin":1959,"end":1963},"obj":"16499990"},{"id":"18982112-17911164-243703","span":{"begin":1985,"end":1989},"obj":"17911164"},{"id":"18982112-17911164-243704","span":{"begin":2787,"end":2791},"obj":"17911164"},{"id":"18982112-15043218-243705","span":{"begin":3030,"end":3034},"obj":"15043218"},{"id":"18982112-17151600-243706","span":{"begin":3049,"end":3053},"obj":"17151600"},{"id":"18982112-16184189-243707","span":{"begin":3239,"end":3243},"obj":"16184189"},{"id":"18982112-16381949-243708","span":{"begin":3266,"end":3270},"obj":"16381949"}],"text":"Atlases for Sharing Gene Expression Data\nMany neuroscientists have been calling for a system for the mouse brain where digital atlases serve as the framework used to traverse the brain and information linked to it (Baldock et al., 2003; Bjaalie, 2002; Boline et al., 2007; MacKenzie-Graham et al., 2003; Martone et al., 2004; Toga, 2002). In contrast to data repositories, which allow simple access to data through a single interface, sophisticated digital atlases backed by the appropriate technology can act as a neuroinformatics hub facilitating access to different databases, information sources, and related documents and annotations. They may act as the scaffold in which otherwise unrelated data may be housed and correlated, providing an intuitive interface to share, visualize, analyze, and mine data of multiple modalities, scales, and dimensions.\nThe semantic and spatial information tied to an atlas can add a dimension to data in a manner that exponentially increases its potential use and reusability. Semantic linking of data to the atlas requires the data provider to register it with an ontology or controlled vocabulary, while spatial registration requires alignment of an image to the atlas. This information is then used to place data into the context of the atlas, allowing it to inherit information tied to the spatial coordinates of the atlas. A perfect example of this is gene expression image data. Spatial information in this type of slice data is key to interpreting results, yet these images lose anatomical context during the data collection process. Thus atlas-based tools for organizing and analyzing this and related types of data may be used to create a system ideal for sharing data.\nSeveral projects offer access to gene expression image data with differing levels of spatial mapping in the mouse nervous system. These projects have been comprehensively reviewed (Brumwell and Curran, 2006; Koester and Insel, 2007; Sunkin, 2006; Sunkin and Hohmann, 2007) and among others, include the Allen Brain Atlas (ABA, www.brain-map.org), BGEM (www.stjudebgem.org), GENSAT (www.gensat.org), GenePaint (www.genepaint.org), EurExpress (www.eurexpress.org), MGI (http://www.informatics.jax.org/) and EMAP/EMAGE (http://genex.hgu.mrc.ac.uk). Several of these projects are geared toward the developmental stages of the mouse. As illustrated by the images from GENSAT and MGI in Figure 1, the ability to examine spatial and temporal expression patterns is crucial for developing correlations between genotype and phenotype as well as for interpreting and comparing findings across experiments. Also illustrated, is that the anatomical information tends to be sparse in these data sources, as is also the case with BGEM, GenePaint, and EurExpress (Sunkin and Hohmann, 2007). ABA and EMAP/EMAGE differ from these sources in that in addition to linking the images to semantic information, they have also linked their images to spatial information by registering their images to a reference atlas (Baldock et al., 2003; Lein et al., 2007). While this extra step can be both difficult and time consuming, it adds the potential for a great deal of analytical power and the ability to generate spatial queries (Carson et al., 2005; Christiansen et al., 2006; Leergaard and Bjaalie, 2007).\nFigure 1 Examples of databases which manage gene expression image without an anatomical framework. GENSAT and MGI both provide a rich repository of image data for gene expression. Neither uses a standard framework for data organization, rather, both describe the pattern and location of gene expression revealed by the image. GENSAT (A) describes the expression pattern for major structures in each image, while the MGI (B) summarizes a batch of assays (the MGI dataset shown here were used to derive the Lef1 gene examples shown in Lee et al., 2007). While these annotations aid interpretation, the ability to query or analyze these data in this format is very limited. As most resources are built for sharing a specific set of data, most discussed above are not yet set up to easily link their data to that offered by other groups, adding a barrier to analyzing data across experiments. Also, with the exception of EMAGE, mechanisms are not readily available for an individual to easily put data into a semantic and spatial framework that facilitates comparison of their own data to others. For these reasons, it still requires a great deal of research and work to compare data collected in different experiments, which is one of our communities greatest desires, but also most difficult tasks. These are some of the recent drivers for the call to create interoperability across data resources and offering access to this system to any scientist, within the context of a digital atlasing framework.\nIntegrating these and other data sources via such a framework would allow a researcher to easily query across these resources. One could look for studies of different collection modalities, strains, developmental stages, or disease models, or examine the expression patterns of genes or regions of interest across multiple studies. For example, a scientist investigating a disease model of Parkinson's with the microarray technique finds that an unexpected gene in the caudate/putamen area seems to have a reverse correlation with motor deficits. Wanting to know more, he uses this system to find data from other experiments that have examined the same disease model. He finds a MRI dataset illustrating a change in the shape and decrease in the volume of that area in later stages of the disease, and a high-resolution confocal dataset from this same region shows abnormal cell morphology with disease progression. In addition, he finds that expression of this gene in this region in normal animals decreases with age and that it is expressed at a higher level throughout life in a different mouse strain which shows resistance to Parkinson's. Compiling and analyzing data from these different experiments, many of which he did not even realize were applicable to his situation, allows him to more fully examine the potential role of this gene and to better inform his next experiments.\nWhile individual researchers often do a similar type of information gathering on their own, it can be difficult to examine other datasets, or we miss a relevant dataset because the data producers published it in relation to a very different topic. As diverse data generation continues to grow at an accelerated rate, we are in dire need of systems that make it easy for our community to contribute, organize, and find relevant data. While we are a long way from a fully implemented system that could perform the previous example, different groups have already created many of its components."}

{"project":"TEST0","denotations":[{"id":"18982112-190-198-243695","span":{"begin":231,"end":235},"obj":"[\"15043218\"]"},{"id":"18982112-205-213-243696","span":{"begin":246,"end":250},"obj":"[\"11967564\"]"},{"id":"18982112-231-239-243697","span":{"begin":298,"end":302},"obj":"[\"15043223\"]"},{"id":"18982112-234-242-243698","span":{"begin":320,"end":324},"obj":"[\"15114360\"]"},{"id":"18982112-235-243-243699","span":{"begin":332,"end":336},"obj":"[\"12374427\"]"},{"id":"18982112-72-80-243700","span":{"begin":1920,"end":1924},"obj":"[\"16728446\"]"},{"id":"18982112-97-105-243701","span":{"begin":1945,"end":1949},"obj":"[\"17521453\"]"},{"id":"18982112-111-119-243702","span":{"begin":1959,"end":1963},"obj":"[\"16499990\"]"},{"id":"18982112-137-145-243703","span":{"begin":1985,"end":1989},"obj":"[\"17911164\"]"},{"id":"18982112-173-181-243704","span":{"begin":2787,"end":2791},"obj":"[\"17911164\"]"},{"id":"18982112-232-240-243705","span":{"begin":3030,"end":3034},"obj":"[\"15043218\"]"},{"id":"18982112-229-237-243706","span":{"begin":3049,"end":3053},"obj":"[\"17151600\"]"},{"id":"18982112-183-191-243707","span":{"begin":3239,"end":3243},"obj":"[\"16184189\"]"},{"id":"18982112-210-218-243708","span":{"begin":3266,"end":3270},"obj":"[\"16381949\"]"}],"text":"Atlases for Sharing Gene Expression Data\nMany neuroscientists have been calling for a system for the mouse brain where digital atlases serve as the framework used to traverse the brain and information linked to it (Baldock et al., 2003; Bjaalie, 2002; Boline et al., 2007; MacKenzie-Graham et al., 2003; Martone et al., 2004; Toga, 2002). In contrast to data repositories, which allow simple access to data through a single interface, sophisticated digital atlases backed by the appropriate technology can act as a neuroinformatics hub facilitating access to different databases, information sources, and related documents and annotations. They may act as the scaffold in which otherwise unrelated data may be housed and correlated, providing an intuitive interface to share, visualize, analyze, and mine data of multiple modalities, scales, and dimensions.\nThe semantic and spatial information tied to an atlas can add a dimension to data in a manner that exponentially increases its potential use and reusability. Semantic linking of data to the atlas requires the data provider to register it with an ontology or controlled vocabulary, while spatial registration requires alignment of an image to the atlas. This information is then used to place data into the context of the atlas, allowing it to inherit information tied to the spatial coordinates of the atlas. A perfect example of this is gene expression image data. Spatial information in this type of slice data is key to interpreting results, yet these images lose anatomical context during the data collection process. Thus atlas-based tools for organizing and analyzing this and related types of data may be used to create a system ideal for sharing data.\nSeveral projects offer access to gene expression image data with differing levels of spatial mapping in the mouse nervous system. These projects have been comprehensively reviewed (Brumwell and Curran, 2006; Koester and Insel, 2007; Sunkin, 2006; Sunkin and Hohmann, 2007) and among others, include the Allen Brain Atlas (ABA, www.brain-map.org), BGEM (www.stjudebgem.org), GENSAT (www.gensat.org), GenePaint (www.genepaint.org), EurExpress (www.eurexpress.org), MGI (http://www.informatics.jax.org/) and EMAP/EMAGE (http://genex.hgu.mrc.ac.uk). Several of these projects are geared toward the developmental stages of the mouse. As illustrated by the images from GENSAT and MGI in Figure 1, the ability to examine spatial and temporal expression patterns is crucial for developing correlations between genotype and phenotype as well as for interpreting and comparing findings across experiments. Also illustrated, is that the anatomical information tends to be sparse in these data sources, as is also the case with BGEM, GenePaint, and EurExpress (Sunkin and Hohmann, 2007). ABA and EMAP/EMAGE differ from these sources in that in addition to linking the images to semantic information, they have also linked their images to spatial information by registering their images to a reference atlas (Baldock et al., 2003; Lein et al., 2007). While this extra step can be both difficult and time consuming, it adds the potential for a great deal of analytical power and the ability to generate spatial queries (Carson et al., 2005; Christiansen et al., 2006; Leergaard and Bjaalie, 2007).\nFigure 1 Examples of databases which manage gene expression image without an anatomical framework. GENSAT and MGI both provide a rich repository of image data for gene expression. Neither uses a standard framework for data organization, rather, both describe the pattern and location of gene expression revealed by the image. GENSAT (A) describes the expression pattern for major structures in each image, while the MGI (B) summarizes a batch of assays (the MGI dataset shown here were used to derive the Lef1 gene examples shown in Lee et al., 2007). While these annotations aid interpretation, the ability to query or analyze these data in this format is very limited. As most resources are built for sharing a specific set of data, most discussed above are not yet set up to easily link their data to that offered by other groups, adding a barrier to analyzing data across experiments. Also, with the exception of EMAGE, mechanisms are not readily available for an individual to easily put data into a semantic and spatial framework that facilitates comparison of their own data to others. For these reasons, it still requires a great deal of research and work to compare data collected in different experiments, which is one of our communities greatest desires, but also most difficult tasks. These are some of the recent drivers for the call to create interoperability across data resources and offering access to this system to any scientist, within the context of a digital atlasing framework.\nIntegrating these and other data sources via such a framework would allow a researcher to easily query across these resources. One could look for studies of different collection modalities, strains, developmental stages, or disease models, or examine the expression patterns of genes or regions of interest across multiple studies. For example, a scientist investigating a disease model of Parkinson's with the microarray technique finds that an unexpected gene in the caudate/putamen area seems to have a reverse correlation with motor deficits. Wanting to know more, he uses this system to find data from other experiments that have examined the same disease model. He finds a MRI dataset illustrating a change in the shape and decrease in the volume of that area in later stages of the disease, and a high-resolution confocal dataset from this same region shows abnormal cell morphology with disease progression. In addition, he finds that expression of this gene in this region in normal animals decreases with age and that it is expressed at a higher level throughout life in a different mouse strain which shows resistance to Parkinson's. Compiling and analyzing data from these different experiments, many of which he did not even realize were applicable to his situation, allows him to more fully examine the potential role of this gene and to better inform his next experiments.\nWhile individual researchers often do a similar type of information gathering on their own, it can be difficult to examine other datasets, or we miss a relevant dataset because the data producers published it in relation to a very different topic. As diverse data generation continues to grow at an accelerated rate, we are in dire need of systems that make it easy for our community to contribute, organize, and find relevant data. While we are a long way from a fully implemented system that could perform the previous example, different groups have already created many of its components."}