Free download. Book file PDF easily for everyone and every device. You can download and read online The Facts on File Illustrated Guide to the Human Body, Brain and Nervous System file PDF Book only if you are registered here. And also you can download or read online all Book PDF file that related with The Facts on File Illustrated Guide to the Human Body, Brain and Nervous System book. Happy reading The Facts on File Illustrated Guide to the Human Body, Brain and Nervous System Bookeveryone. Download file Free Book PDF The Facts on File Illustrated Guide to the Human Body, Brain and Nervous System at Complete PDF Library. This Book have some digital formats such us :paperbook, ebook, kindle, epub, fb2 and another formats. Here is The CompletePDF Book Library. It's free to register here to get Book file PDF The Facts on File Illustrated Guide to the Human Body, Brain and Nervous System Pocket Guide.
Most frequently terms

Motor neurons, located in the central nervous system or in peripheral ganglia, transmit signals to activate the muscles or glands. The brain's connections and thinking ability grew over thousands of years of evolution. This code packages up genetic information and sends it from nerve cells to other nearby nerve cells, a very important process in the brain. There are a number of tests and procedures to diagnose conditions involving the nervous system.

In addition to the traditional X-ray, a specialized X-ray called a fluoroscopy examines the body in motion, such as blood flowing through arteries, according to the NIH. Positron emission tomography PET is a procedure that measures cell or tissue metabolism and brain activity to detect tumors or diseased tissue or tumors, the NIH noted. A spinal tap places a needle into the spinal canal to drain a small amount of cerebral spinal fluid that is tested for infection or other abnormalities, according to the NIH.

Mayo Clinic also noted that the nervous system can also be affected by vascular disorders such as:. Infections such as meningitis, encephalitis, polio, and epidural abscess can also affect the nervous system, the NIH noted. Treatments vary from anti-inflammatory medications and pain medications such as opiates, to implanted nerve stimulators and wearable devices, Gozani said.

The branch of medicine that studies and treats the nervous system is called neurology, and doctors who practice in this field of medicine are called neurologists. There are also physiatrists, who are physicians who work to rehabilitate patients who have experienced disease or injury to their nervous systems that impact their ability to function, according to the ABPN.

Fig 3B depicts the distribution of BRO scores for three gene families involved in regional specificity during brain development: axon guidance genes, Hox genes, and Pax genes. Genes involved in early brain developmental have been shown to have regional expression patterns in the adult [ 39 , 40 ][ 23 ]. This suggests that beyond their embryonic role, genes involved in axon guidance may assume other functional roles in the adult brain. Second, Hox genes play a major role in anterior-posterior patterning across the body and across the brain during development and largely retain these patterns in the adult body [ 16 ].

Their role in the adult brain is less clear. This suggests that unlike other gene groups discussed above, Hox genes are less involved in regional patterns in the adult brain. These view is also supported by Takahashi et al. Finally, we examined Pax genes. These genes are involved in early regionalization of the embryo brain and were suggested to play a role in differentiation and maintenance of specific subsets of cells in the adult brain [ 41 , 42 ]. It has been shown before that genes important for brain developmental have regional expression patterns in the adult [ 39 , 40 ], including genes involved in brain connectivity [ 23 ].

Interestingly, PAX6 is a major determinant of regionalization in the mammalian brain [ 9 , 10 , 42 ]. It was shown to be essential to cortex development, to mark cortex regionalization and to regulate radial migration of neuronal precursors [ 39 , 43 , 44 ]. The differential areal expression pattern of PAX6 in the adult raises the hypothesis that PAX6 continues to play a region-specific role in the adult brain. The above results suggest that spatial regionalization of human brain expression is present both in brain-specific functions and also in more generic processes that can be found in simpler organisms.

Importantly, spatial regionalization of the nervous system is not unique to mammals or vertebrae, and some of the mechanisms controlling spatial patterning are shared across evolutionary-remote species [ 45 ]. For instance, Hox genes, whose expression exhibit anterior-posterior gradients in mammals, also hold spatial information in species that diverged from the human lineage early in evolution [ 46 — 48 ]. The natural question therefore arises: how is brain regionalization of a gene related to the evolutionary age of that gene?

For instance, one may hypothesize that genes with high BRO agreement would be genes those that evolved recently, in organisms having a nervous system similar to the mammalian brain. To test this hypothesis, we compare the BRO index with an index quantifying the evolutionary age genes [ 49 ]. These older genes are also active in signaling pathways and other basic functions in the cell S4 Table.

The top BRO-scoring genes have orthologs across a wide variety of species, and participate in functions that are not specific to neural processes. Presumably, these genes were conserved as the result of a pressure to preserve these basic functions. For example, the gene ENC1 encodes an actin-binding protein involved in regulation of neuronal process formation and in differentiation of neural crest cells. On the other range of the evolutionary timeline, genes associated with speciation of primates obtain lower BRO scores on average.

These results suggest that genes with strong BRO scores and spatial patterns are not necessarily specific to neural processes, but rather that the brain spatially tunes the expression of genes involved in fundamental molecular functions. With that said, it is also possible that these newer genes exhibit more refined differences across brain regions, but that these changes are not captured by the current coarse-scale analysis also compare with [ 50 ]. A The median BRO score as a function of the evolutionary age of genes.

Older genes receive on average higher BRO scores than evolutionary recent genes. B Focus on the distribution of BRO scores for the oldest gene group and of the most recent primates gene group. Genes in the cellular organisim group have a median BRO-score of 0. The two distributions are signifatcly different. Expression variability has many contributing factors, including subject-to-subject variability, regional variability and experimental noise.


  1. Henry More. The Immortality of the Soul.
  2. Easy Mac OS X Lion (2nd Edition)?
  3. kindle电子书下载代找书籍;

The above results suggest that the variability between brain regions is significant for most genes. But, how large is regional variability compared to other sources of expression variability? To answer this question, we used principal component analysis PCA to extract the main axes of variability in the data see Methods. Interestingly, the PCA of the human expression was also analyzed previously by Tan et al.

Tan et al. In that space, they found that neurons and oligodendrocytes are on the opposite end of the first principal component. Here we address the complementary analysis, looking for the dimensions that preserve the sample-to-sample variability. Samples are colored by the brain region from which they were taken. Brain regions are well separated in this projection, in a way that matches the anterior-posterior axis and the BRO.

The isolated cluster of samples on the left belongs to the cerebellum, which is well known to exhibit a unique molecular and cellular organization [ 21 , 51 , 52 ]. This analysis shows that the BRO is a major determinant of variability in human brain transcriptome. A The samples are colored according to the position of the corresponding embryonic region, using the same color scheme as in Fig 1A. B The samples are colored according to one of the six donors. As a comparison, Fig 5B shows the same projection on the two top PCs, but this time the samples are colored by the subject from which each sample was taken.

Expression differences between people are pronounced mostly in frontal regions top right samples , but are dramatically weaker than the differences between brain regions. Subject-to-subject differences are more pronounced when projecting on the 3 rd and 4 th principal components S2 Fig. To quantify the relative contribution of subject identity and region of origins to expression variability, we computed the fraction of variance explained by these two features.

For every gene, we examined it expression across samples separately and computed the fraction of explained-variance See Methods S8A Fig.

Circulatory System

The subject-identity explains 0. Together, both sources explain nearly half of the sample variability median at 0. Using the same dataset, Hawrylycz et al. They showed that genes with conserved patterning across subjects display strong relationships to anatomical structure, functional connectivity and other features of the human brain. With many genes exhibiting spatial expression that matches the developmental origin of brain regions, the question remains if and how expression variability is used by the brain to tune the functional properties of cells and circuits.

One particularly interesting aspect of such tuning is how the brain controls the expression of similar genes, including paralog and other functionally-related gene pairs. In many cases, the brain is known to switch from expressing one paralog variant to another variant. These developmental switches can be traced to occur within a brain region, and in some cases well after birth [ 54 , 56 ].

Here we study spatial switching in pairs of genes, where genes coding for different protein variants are expressed in different brain regions. We set to study the relation between spatial expression switching and developmental origin of regions, using the per-gene BRO-agreement score. To study fine spatial tuning, we aimed to focus on pairs of genes that share similar functions. To collect such gene pairs we used two approaches.

Diagnosing nervous system conditions

First, we used a set of paralog genes defined by Ensembl denoted ensemble-based paralogs. Second, to further focus on genes with putative similar function, we collected pairs of genes that share the same functional role in cellular pathways, as captured by KEGG. We also required that these gene pairs have a significant sequence similarity and denote this set Kegg-based pairs see Methods.

Both sets were restricted to brain-related synaptic pathways. We first compared the distribution of spatial correlation strengths of similar gene pairs, quantified by log p-values on ABA data. We found that both the KEGG-based gene pairs and the Ensembl-based paralogs are significantly more spatially anticorrelated than random gene pairs Fig 6A. The spatial correlations of KEGG-based pairs are fairly consistent when compared to the correlation using the same gene pairs in adult brains from the Kang dataset, considering they were measured by different labs and in different brain regions Fig 6B.

A Distribution of gene pairs with anti-correlated spatial expression. Ensembl-based paralogs include pairs of paralogs as defined by Ensembl where both genes in a pair are included in one of same 17 pathways see Methods. Baseline corresponds to the distribution expected at random. Each point correspond to the median correlation across adult subjects, in one gene pair total of pairs.

C-F Examples of development of spatial correlations in the Serotonin system. C The pair of genes coding for Serotonin receptors HTR2A and HTR1F exhibit a continuous rise in spatial correlation, riding from slightly negative in early embryonic development to strong positive correlation. D The pair of genes coding for Serotonin receptors HTR5A and HTR2C show a sharp transition from positive to negative spatial correlation in early development, which is then preserved through life.

One interpretation of these findings is that the brain tunes the expression of pairs of functionally-related pairs of genes, such that they are expressed differently in brain regions, and that this tuning is in strong agreement with the developmental origin. Together with the BRO results, these findings suggest that correlated spatial expression may be formed early in development.

To test this hypothesis we computed the spatial correlations for each subject in the Kang data, which allows tracing how spatial correlations develop with age. Far fewer pairs of Ensembl paralogs exhibit a significant trend 6. Interestingly, their spatial correlation is negative prenatally and is around zero around birth. However, it continuously grows throughout life, reaching high positive correlation at adulthood.

This pattern is interesting for several reasons. First, the gradual increase in correlations throughout life is not likely to be caused by changes in cell proportions, since there is a significant change in correlation between childhood and adulthood. Second, other genes in the Serotonin system exhibit different patterns. Here the early embryonic positive correlation is replaced by a negative correlation around birth, which remains quite stable during life.

HTR5A , two serotonin receptors. HTR7 is known to be involved in both early and post-natal development [ 58 ]. One possible interpretation of the prevalence of high BRO-agreement scores is that the expression patterns of many are determined early in development, and are preserved through life and in the adult brain.

Alternatively, it is also possible that gene-expression changes in a dynamic way through life, but keep following patterns that agree with the embryonic origin of regions. To test these two hypotheses, we quantified the relation between the strength of expression changes of pairs through life, and the BRO scores of the gene pair. These results are consistent with the view that spatial expression patterns in the adult are not a mere reflection of the brain structure as determined in early development, but are tuned to use genes coding for different protein variants in a differential way across the brain.


  1. Capital: A Critique of Political Economy, Volume 3;
  2. FreeMedicalBooks – Nursing – History of Medicine!
  3. Psychoanalytic Psychotherapy in the Kleinian Tradition.
  4. Books for Middle School Students.

To characterize the areal patterns of gene expression in the human brain, we analyzed two datasets of mRNA expression from post-mortem adult donors. For each gene, we computed an index that measures how its expression pattern agrees with a hierarchical ontology of brain-regions, based on their developmental origin. The fact that such a large fraction of the human genome is differentially expressed across brain regions suggests that control of expression in the brain is largely region-specific.

When focusing on genes that are expressed specifically in neurons, glia and oligodendrocytes, we find that cell-type specific genes tend to strongly agree with the tree-structured ontology. This suggests that not only do these markers differ between regions, as suggested by Ko et al. Interestingly, significant BRO scores are not limited to neurons, which are often known to differ across brain regions, but are also observed in glia-specific genes, which are often viewed as performing brain-wide and generic functions.

Having adult expression patterns that strongly agree with the developmental brain region ontology could have various interpretations. First, adult spatial expression patterns could be determined by the embryonic origin of each region, for example because brain regions differ by their cell-type profiles, or due to the expression of region-specific markers. Alternatively, adult expression may reflect delicate tuning of expression where different brain regions utilize different protein variants, optimized for the function of each brain region.

We find evidence that support the second alternative. First, gene with high BRO-scores tend to change their expression more during development. Second, pairs of functionally-related genes participating in a similar role in synaptic pathways have stronger spatial anti-correlation than paralogs in those pathways. Finally, in those pairs of functionally related genes, pairs with higher BRO scores tend to have stronger spatial anti-correlation.

The approach we presented has various limitations. Transcriptome data measured from brain tissues involves a mixture of various cell types whose proportions and conditions are not known. Developing demixing approach to infer proportions from the mixture [ 25 ] is an important challenge, can be based on single-cell transcriptomics as in Darmanis et al , and is likely to significantly change our understanding of brain transcriptome.

Genes involved in patterning and axon guidance clearly exhibit regional patterns during early development [ 59 ]. The above results show that their expression continues to be governed by the region ontology in the adult brain, long after their developmental role has been completed. As one specific example, consider a gene from the top BRO-scorers in the ABA dataset, FEZF2 forebrain embryonic zinc finger protein 2 , a transcription repressor involved in specification of subcerebral projection neurons [ 60 , 61 ]. FEZF2 is believed to play a role in layer and neuronal patterning of subcortical projections and axonal fasciculation and was shown to regulate axon targeting of layer 5 subcortical projection neurons, where axons of FEZF2 deficient mice failed to reach their targets [ 62 ].

In the adult human brain, our results show that FEZF2 retains strong areal differences in adulthood, and is strongly expressed in the cortex, less so in the midbrain and the least in the hindbrain S6 Fig. Indeed, the mouse variant of FEZF2 is known to be expressed in adult projection neurons [ 62 ]. Importantly, these results suggest that FEZF 2 assumes another functional role in the adult cortex. Indeed, it has been shown that projecting neurons in the mouse motor cortex expressing Fezf2 have distinct physiological characteristics [ 63 ]. The abundance of genes that retain their areal differential expression in adulthood suggests that this may be the general case, and many genes that play a role in brain development later assume new roles that affect the function of the adult brain.

The fraction of genes having distinct areal expression pattern has been previously estimated using a different method ANOVA. In the Kang data, it was found to be on the order of hundreds of genes in the adult human brain Pletikos et al. ANOVA estimates are sensitive to differences in the mean expression of regions, regardless of the region ontology, and could capture genes whose expression pattern in some brain regions is different from others. As such, they are more sensitive to genes that are uniquely expressed in one or few region.

The BRO-score can therefore be viewed as a complementing measure, which is sensitive to global areal-differential expression that is consistent with the brain region ontology. The fact that the expression of most genes in the adult brain is governed by earlier development stages suggest that many studies which deal with regional differential expression should be carefully interpreted. For example, combining samples taken from ontology-distant regions would lead to large expression variance, reflecting the developmental origin of the structures tested.

Furthermore, areal differential expression should be measured compared to a baseline expression profile that takes in to account the region ontology. The results in this paper suggest that spatial expression patterns in the adult human brain are controlled in a way that follows the embryonic origin of regions, but at the same time that spatial patterns of related genes may change during development in a correlated way.

It remains to be discovered which transcription control mechanisms maintain these distinct areal expression patterns. We analyzed gene expression data from two sources. We mapped microarray probes to genes based on mapping provided by the Allen Institute. When multiple probes were available for a gene, we selected the probe that was most consistent across the 6 human subjects; suggested by human.

Specifically, when we analyzed genes with multiple probes, we first computed the expression correlation across regions of each probe, then averaged the correlation scores across all pairs of subjects and chose the probe that was most correlative. Overall, we analyzed transcripts. The number of samples per donor ranged from to , for a total of tissue samples. The second dataset was a set of microarray samples collected by Kang and colleagues from 57 postmortem brains containing expression values for genes [ 6 ].

We refer to this dataset as Kang We limited the analysis to donors that are older than 12, yielding a total of 20 donors and tissue samples. For this analysis we also used the pre-natal subjects and early-childhood subjects. The subject ages range from 13 post conception weeks to 82 years. We grouped the subjects into 12 age-groups following the original Kang paper, and computed BRO scores per age group.

We used the brain region ontology hierarchy provided by the Allen institute human. From the full set of regions we used two ontologies: A fine region-ontology with regions which had measurements that were associated with them, and a coarse region-ontology with 16 brain regions. The list of 16 gross regions is given in supplemental S2 Table. The coarse part of the ontology upper part in Fig 1A was largely based on brain development while the fine parcellation of regions was based more on cytoarchitecture. The results we report are based on the coarse 16 region-ontology.

We report below results for both coarse and fine grained ontologies. Measurements from the Kang dataset were obtained from 16 regions. We mapped those regions to 16 regions of the Allen ontology, and the mapping is given in supplemental S3 Table. The BRO-agreement score was computed separately for each gene as follows.

For each pair of samples a , b , we define their tree similarity as the distance number of edges between the regions in the ontology hierarchy tree d tree a , b. We define their expression similarity as the absolute difference between the expression values of the two samples for the current gene i — d expression a i , b i. We computed the two distances over all pairs of tissue samples, and computed the Spearman correlation between the two as the BRO score. To generate random scores, we calculate Bro-agreement scores of permutated vectors. We also tested a second ontology-agreement score based on triplet ranking.

We randomly selected 10 6 sample triplets a, b, c and computed the fraction of times that a triplet is ranked with the same ordering in both the tree and the expression distance measures:. To handle biases that could arise from different scales in the samples we also checked a normalized version of the Bro-agreement. In this normalized version we first normalized each sample to zero mean and unit variance and then computed the BRO-agreement. The results were robust to this change and we choose to present the un-normalized BRO-agreement scores.

To handle biases that could arise from the number of regions in the ontology, we used two different granularities of the ontology tree. The first uses 16 gross regions and the second uses the entire tree regions. We tested two ways to combine expression measures from multiple subjects into a single BRO score. First, we simply aggregated all samples of all subjects from a given region, and computed the BRO agreement score.

Frontiers | The Blue-Collar Brain | Physiology

Second, in the ABA dataset, the number of samples per subject is large enough, such that a BRO score can be computed separately for each subject. For consistency with the Kang data, the figures use the first method. We computed BRO-scores which uses the fine region-ontology.

The upper branches in this ontology are more developmental oriented and the lower branches are more cytoarchitecture-driven. The BRO scores of the fine region-ontology and of the coarse 16 region-ontology are very similar with a spearman correlation of 0. To test if these results are sensitive to the number of region available in ontology tree, we repeated the analysis of the ABA dataset at a coarser resolution of 16 regions.

The percent of BRO-significant was tested for robustness using the ABA dataset, where hundreds of samples are available for each subject. For each gene, we calculated its BRO-score but this time for each subject separately. The fraction of significant genes at the group level is slightly higher; suggesting that the groups score manages to remove some of the inter-subject noise.

Expression variability of a single gene across regions is sometimes captured by comparing the mean expression level in each region using ANOVA [ 5 , 6 ]. This approach would find a gene as significant even if it is differentially expressed in a single region, or if it is expressed in a set of regions regardless of their position in the ontology tree. Hence in principle, the BRO agreement index poses a stronger requirement of agreement with the ontology.

The p -values reported are under the null hypothesis that samples are drawn from regions which have the same mean expression. We then corrected the p -values for multiple comparisons using FDR. For the genome wide analysis of BRO scores We used the human orthologs of the set of genes characterized by Cahoy et al. For testing how spatial variability could be explained by cell-type specific markers, we used a set of known markers collected from various sources including [ 36 ].

To compute the dependence of spatial variability on those markers we fitted a quadratic function for each of the genes separately using a least square loss, and computed the explained variance R 2 , compared to a constant model. For the first set, we used the paralogs available from Ensembl ensembl.

Specifically, these included 17 pathways with KEGG accession numbers , , , , , , , , , , , , , , , , Second, for KEGG-based gene pairs, we created a set of gene pairs designed to capture functionally-related genes, by collecting gene pairs that reside within the same functional element in KEGG pathway repository. These KEGG elements group together proteins with common functionally and interaction partners.

We found these KEGG elements to be usually more functionally-coherent than protein families, and at the same time less specific than protein sub-families. To find pairs whose spatial correlation has a trend, we fitted a linear regression model with least square loss with age as the predicting variable and spatial correlation as the predicted variable. Significance of the trend was measured based on the F statistic of the explained variance and was FDR corrected for multiple hypotheses. We used the set of genes that belong to the human axon guidance pathway. The set was manually curated by KEGG www.

A set of 11 genes collected by Eisenberg et al. We used agglomerative hierarchical with average linkage and Euclidean distance over samples from ABA obtained from six subjects. Samples from the same brain regions were first averaged to create a single profile for each region. We used all samples from ABA to compute the covariance matrix of gene expression levels, and then computed the top principal component of the expression covariance matrix. For each gene, we evaluated how much of its expression variance over samples can be explained using two sources of information: The region the sample was taken from and the identity of the subject the sample was extracted from.

The explained variance was computed for each gene by fitting linear model using each of these sources of information, and using both of them together. Sensitivity to the number of regions in the ontology tree depth was assessed by using two ontologies with different resolutions: A fine-resolution ontology that contained regions and a coarse-resolution ontology of 16 gross regions.

Sensitivity to the number of available samples was assessed by computing the BRO score using random subsets of samples. Enrichment was computed using mGH [ 35 ] testing for enriched biological processes using the full ranked list of BRO-score genes. All top processes are brain-related, with high enrichment for cell signaling and neural development. Pearson correlation between expression vectors of all pairs of tissue samples. Samples are ordered first by gross region than by donor. Samples from regions that are close in the developmental region ontology are highly correlated in their expression profile.

Within each region, samples are also correlated with other samples from the same donor. A Each point is a tissue sample. Samples are colored based on the position of the corresponding embryonic region. B Colors correpsond to donor identity. A significant fraction of the sample variance across the 3 rd and 4 th principal components is explained by subject-to-subject variablity. Color scheme and x-axis scale matche those of Fig 2. A A scatter plot showing BRO-agreement scores for the two datasets. Each light-grey dots corresponds to a single genes a total of 17K genes. The large fraction of BRO-significant genes observed in ABA was not found in the Kang dataset, where the two distributions largely overlap.

Color scheme and x-axis scale match those of Fig 3. Oligoodendrocyte genes are in less agreement with region-ontology than the full set of genes. Hox genes are less in agreement with region-ontology than the full set of genes still significantly larger than random. FEZF2 shows a clear transtion of its expression levels. The samples from the cortex show high experssion values where the samples midbrain has less and the samples of the hindbrain has the least expression.

See a Problem?

A The mean expression levels of the FEZF2 within different region in the human brain the color scheme as consitant with that of Fig 1A. The scatter shows a transiation from high expression levels in the cortex to lower expression levels in the inner brain structures and to the hindbrain.

The Facts on File Illustrated Guide to the Human Body, Brain and Nervous System

The joint information of subject id and region id explains almost half of the sample variance. For each gene, we represented the identity information as 1-hot-vectors, and computed the explained sample-variance by fitting a linear model A Distributions of explained variance across genes, as explained by region, subject or both. B The joint distribution of both the explained variance from region and from subject identity. Trend significance of change in spatial correlation through life was quantified using the standard F-test comparing residual of a linear regression model with a constant model.

We found that BRO-agreement scores are weakly but significantly positively correlated with having a significant trend. Each dot corresponds to one brain related sequence-similar pair. B Similarly, we found that the developmental trend of paralog pairs which are brain related is also positively correlated with the BRO score.

A The Allen ontology used for the human analysis. B The Allen ontology used for the mouse analysis. Both follow ontologies first devide the brain into the 5 embrionic vesiceles and then go into more detailed regionalization. The leaf regions are not the same since the data was gathered in different experiments each with a unique focus.

A mark per gene which indicates it involvment in one of these 11 classes: cell-cell signaling, synaptic transmission, neuron differentiation, neuron projection development, generation of neurons, axon development, neuron development, schizophrenia, autistic disorder, seizures,epilepsy, substance related disorders. The authors are grateful to the Allen Institute for Brain Science for making their data available to the scientific community and for helping us with any questions.

They are also grateful to Noa Liscovitch for valuable discussions and insights. Conceived and designed the experiments: LK GC. Performed the experiments: LK GC. Analyzed the data: LK GC. Wrote the paper: LK GC. Abstract Anatomical substructures of the human brain have characteristic cell-types, connectivity and local circuitry, which are reflected in area-specific transcriptome signatures, but the principles governing area-specific transcription and their relation to brain development are still being studied. Author Summary Genome-wide measurements of gene expression across the human brain can reveal new principles of brain organization and function.

Introduction The human brain is organized in a hierarchy of multiple substructures, whose cell composition and circuitry are believed to allow each substructure to carry out its distinct function. Download: PPT. Results To characterize distinct gene expression patterns across the adult human brain, we analyzed genome-wide expression measurements from two sources. Fig 2. Distribution of BRO-agreement scores of individual genes. Fig 3. The distribution of BRO-agreement scores on different subsets of genes. Evolutionary gene age and BRO-agreement The above results suggest that spatial regionalization of human brain expression is present both in brain-specific functions and also in more generic processes that can be found in simpler organisms.

Fig 4. Evolutionary-older genes have on average higher BRO-scores. Source of spatial variability in expression Expression variability has many contributing factors, including subject-to-subject variability, regional variability and experimental noise. Fig 5.

Projection of the samples from the human6 dataset on the 1st and 2nd principal componenets with two coloring schemes. BRO agreement and spatial variability in paralogs and functionally-related genes With many genes exhibiting spatial expression that matches the developmental origin of brain regions, the question remains if and how expression variability is used by the brain to tune the functional properties of cells and circuits. Discussion To characterize the areal patterns of gene expression in the human brain, we analyzed two datasets of mRNA expression from post-mortem adult donors.

Methods Gene expression measurements We analyzed gene expression data from two sources. Brain region ontology We used the brain region ontology hierarchy provided by the Allen institute human. We randomly selected 10 6 sample triplets a, b, c and computed the fraction of times that a triplet is ranked with the same ordering in both the tree and the expression distance measures: This score gave similar results which are not shown here. Combining scores from multiple subjects We tested two ways to combine expression measures from multiple subjects into a single BRO score.

Robustness across subjects The percent of BRO-significant was tested for robustness using the ABA dataset, where hundreds of samples are available for each subject. Cell type-specific markers For the genome wide analysis of BRO scores We used the human orthologs of the set of genes characterized by Cahoy et al.

Trends in spatial correlation To find pairs whose spatial correlation has a trend, we fitted a linear regression model with least square loss with age as the predicting variable and spatial correlation as the predicted variable. Hox, Pax and axon guidance genes We used the set of genes that belong to the human axon guidance pathway. Highly stable gene A set of 11 genes collected by Eisenberg et al.

Hierarchical clustering analysis We used agglomerative hierarchical with average linkage and Euclidean distance over samples from ABA obtained from six subjects. Principal component analysis We used all samples from ABA to compute the covariance matrix of gene expression levels, and then computed the top principal component of the expression covariance matrix. Explained sample-variance analysis For each gene, we evaluated how much of its expression variance over samples can be explained using two sources of information: The region the sample was taken from and the identity of the subject the sample was extracted from.

Supporting Information. S1 Table. S2 Table. Names of brain regions and their corresponding symbols ABA ontology [ 31 ]. S3 Table. Region notations. S4 Table. S1 Fig. Correlation between tissues.