"... In PAGE 2: ... 3. Methods Our computationally minimalistic model of hippocampal CA3 consists of a sparse- ly connected recurrent network of McCulloch}Pitts neurons ( Table1 ; for details, see [5]). The external inputs represent signals from the entorhinal cortex/dentate gyrus to CA3.... ..."

"... In PAGE 7: ...ata of Sandberg et al. [1]. Our feature-selection strategy, which we call template matching , is depicted schematically in Figure 1. An overview of the results for both the ANOVA and template-matching methods is given in Table1 , which shows the number of genes identified by each method at three different p-value cutoffs. These cutoffs were selected to represent varying levels of false positive risk (described in the Methods section).... In PAGE 7: ... The middle threshold of p lt;10-5 is a value we use as a compromise for the purpose of comparing our results to those of the previous studies. Table1 also shows the number of genes identified in the work of Sandberg et al. A more detailed view of part of the data is provided in Figure 2, which enumerates some genes identified by both methods, ranked by template-match p- value.... In PAGE 7: ...alue. Some (but not all) genes identified by Sandberg et al. [1] and Sandberg [12] are also given high scores (low p-values) by our methods, as indicated in Figure 2. As indicated in Table1 and Figure 2, even at conservative p-value cutoffs we identified additional genes not noted by Sandberg et al. [1].... In PAGE 8: ...Paul Pavlidis Page 8 7/30/2001 though ANOVA has a tendency to give more low p-values and there are some significant outliers. In agreement with the trend illustrated in Figure 3, the number of genes that make each cutoff in Table1 are very similar for strain-specificity. However, the template-matching method tends to give low p-values to many more genes on the basis of region-specificity.... In PAGE 8: ... However, the template-matching method tends to give low p-values to many more genes on the basis of region-specificity. For all tissue effects, at a p-value of 10-5, the template method identifies more than twice as many genes as ANOVA, and at less conservative cutoffs the difference is even greater ( Table1 ). An example of a discrepancy is synuclein (GenBank C79089), which had a p-value of 9.... In PAGE 9: ... [1] also identified an additional 49 genes which varied between strains for particular regions. We have not specifically searched for such genes; the ANOVA interaction effects would correspond most closely to such a category, but very few genes were found to have significant interaction effects ( Table1 ). Preliminary results (not shown; see supplemental data) using templates which select specifically for genes expressed in one region in one strain suggest that this strategy will be useful for identifying more complex expression patterns as well as the simple ones we describe in detail here.... In PAGE 10: ...Paul Pavlidis Page 10 7/30/2001 with respect to brain region ( Table1 ), or ~6% of the total, compared to the 242 region-specific genes identified by Sandberg [12] and Sandberg et al. [1].... In PAGE 10: ... The bulk of genes showing region effects identified appear to be due to cerebellum enrichment or depletion (339/638 at template-match p-values less than 10-5). The midbrain was the second-most distinct single region ( Table1 ). In contrast, the cortex had the least distinct expression pattern, being very similar to other forebrain regions, as pointed out by Sandberg et al.... In PAGE 10: ... In contrast, the cortex had the least distinct expression pattern, being very similar to other forebrain regions, as pointed out by Sandberg et al. When considered together, the forebrain (represented by cortex, amygdala, hippocampus and entorhinal cortex) has many genes which distinguish it from the midbrain and cerebellum ( Table1 ). Taken separately, the cortex, amygdala, hippocampus and entorhinal cortex had relatively few genes with distinct patterns (Table 1 and Figure 2).... In PAGE 10: ... When considered together, the forebrain (represented by cortex, amygdala, hippocampus and entorhinal cortex) has many genes which distinguish it from the midbrain and cerebellum (Table 1). Taken separately, the cortex, amygdala, hippocampus and entorhinal cortex had relatively few genes with distinct patterns ( Table1 and Figure 2). The single amygdala- specific gene (ARP-1) identified by Sandberg et al.... In PAGE 12: ...l. do not rank particularly high in our results. Furthermore, in some cases, such as for the entorhinal cortex at more conservative p-value cutoffs, we select fewer genes than Sandberg et al. even though at the same cutoff for other regions we select more ( Table1 ). Thus, our increase in sensitivity does not come at the cost of selectivity (ability to avoid spurious positive findings) compared to the ad hoc approach.... In PAGE 29: ...Paul Pavlidis Page 29 7/30/2001 Figure legends Table1 : Summary of results. The number of genes identified by each method, and by Sandberg et al.... ..."

"... In PAGE 9: ..., 1993; Miller and Desimone, 1994). The relative incidences of visual responsiveness, stimulus selec- tivity, sample-selective delay activity, match enhancement, and match suppression are given in Table1 . All of the values for the Figure 9.... ..."

"... In PAGE 10: ... (3) Given that most criterion 2. Previous studies based on cRNA probes showed a relatively oligonucleotides used were 45-mers with similar GC content (Table ,I), widespread distribution of this subunit mRNA in adult rats, notably high each oligonucleotide probe constituted a control for the specificity of levels in the cerebral cortex layer IV, entorhinal cortex layer II, anterior the others. (4) The addition to the hybridization mixture of a 20-fold and ventral thalamic nuclei, medial and lateral geniculate nuclei, medial Figure 6.... In PAGE 13: ... From El2 to E13, cr3, (~4, and l32 mRNA levels strikingly increased in the brain. On the other hand, l34 mRNA levels peaked at El2 and were already reduced at El3 (Table 2, Figs. 4-9).... In PAGE 14: ... 8). At E15, the general pattern of labeling of 014 and l32 mRNAs diverged from that of 013 and l34 (Table 2, Figs. 10-13).... In PAGE 14: ..., Figs. 14-17). p2 remained diffusely dis- tributed throughout the brain. 014 displayed a relatively more heterogeneous distribution, being absent in the basal ganglia and olfactory bulb and highly concentrated in the ventromedial hy- pothalamic nucleus, several thalamic nuclei and a few brainstem nuclei (Table 2). a3 and l34 were highly expressed in the medial habenula (Figs.... In PAGE 18: ... 14, 24), was observed. Overall, during prenatal and perinatal development the pat- terns of nAChR subunit expression seem specific for the differ- ent categories of ganglia (Table 7). In the somatic and gustatory sensory ganglia (trigeminal, geniculate, and superior glosso- pharyngeal ganglia) the initial expression of all nAChR subunits studied is followed by a restriction to only the p2 transcript.... In PAGE 24: ... In fact, in vitro (w4-B2 or 013~B4 containing recep- tors have relatively high affinity for nicotine, whereas 013~B2 combination has the lowest affinity for nicotine, though rela- tively high affinity for ACh (Luetje and Patrick, 1991). Starting from late prenatal development, the combinations of nAChR subunits appear structure specific (Table 7). The analysis of other nicotinic subunits not considered here should further increase the number of region-specific isoforms.... ..."

"... In PAGE 6: ... In the S1 and Au1 regions the principal axes coincided with the laboratory coordinates. Table1 shows that in the S1 re- gion, ADCW was significantly higher (0.75 H11006 0.... In PAGE 9: ... In such acute pathologies a good correlation between changes in ADCTMA, ADCW, and ECS volume has been found (32). Both the TMA and MR measurements in control animals revealed anisotropy in the primary somatosensory and auditory cortices (see Table1 and Fig. 4).... In PAGE 9: ... We did not find changes in ADCW anisotropy in cortical re- gions S1, S2, or Au1. However, ADCTMA anisotropy was weak in the S1 area when compared to the auditory cortex (see Table1 ). This divergence in the results of the two methods is probably caused by the different diffusion be- havior of TMAH11001 and water in brain tissue.... ..."