A Transcriptomic Atlas of Mouse Neocortical Layers

Online Supplementary Tables

Back to main download page

AB1B2CDEF
mt-Cytbmt-Cytbmt-Cytbmt-Cytbmt-Cytbmt-CytbMbp
mt-Co1mt-Nd1mt-Nd1mt-Nd1mt-Nd1mt-Nd1mt-Cytb
mt-Nd1Nrgnmt_Rnr2mt-Rnr2mt-Rnr2mt-Rnr2Plp1
mt-Rnr2mt-Rnr2mt-Co1mt-Rnr1mt-Co1Mbpmt-Nd1
mt-Rnr1mt-Rnr1mt-Nd5mt-Nd6mt-Rnr1mt-Co1Fth1
mt-Nd4mt-Nd2mt-Nd6mt-Nd5mt-Nd5mt-Rnr1mt-Co1
Nrgnmt-Nd6mt-Nd4mt-Co1mt-Nd2Snap25mt-Rnr2
mt-Nd5mt-Nd4mt-Nd2mt-Nd2mt-Nd6NrgnGm10927
mt-Co2mt-Nd5mt-Rnr1mt-Nd4mt-Nd4mt-Nd5Mobp
mt-Nd6mt-Co1NrgnSnap25Snap25mt-Nd2mt-Nd5
Online Supplementary Table 1. Ten most highly expressed known genes in each sample. FPKM values of these Ensembl genes (release 57) ranged from 11,532 to 1,291 - nearly one order of magnitude. Most genes are mitochondrially encoded. This is unlikely to be an effect of mtDNA contamination, since the read coverage on chrM strongly stacks up in exons. MGI translations of Ensembl gene IDs are given.

AB1B2CDEF
A1.000
B10.9211.000
B20.9420.9651.000
C0.9220.9680.9721.000
D0.9190.9650.9670.9791.000
E0.9150.9570.9620.9730.9781.000
F0.9070.9500.9500.9610.9660.9721.000
Online Supplementary Table 2. Inter-sample expression Spearman rank correlations. These correlations use the FPKM values of Ensembl (release 57) genes as quantified by cufflinks. The table is, of course, symmetrical about the diagonal.

AUCprecisionrecallgenes predictedgenes expected
Layers 2/30.7649%54%226205
Layer 40.8736%63%6738
Layer 50.7667%68%201198
Layer 60.8447%51%6964
Layer 6b0.8154%60%125113
Unpatterned0.6739%42%241224
Online Supplementary Table 3. Classifier statistics for patterning of neuronal genes. Three statistics from each classifier are presented: Area Under the receiver operating characteristic Curve (AUC), proportion of the “genes predicted” that truly have elevated expression in the layer (precision), and the proportion of genes with elevated expression in the layer that are contained in the “genes predicted” (recall). The next column contains the number of genes predicted by each classifier out of 640 classifiable neuron-specific genes. The last column represents the number of genes expected to be preferentially expressed in this layer, and is calculated as [(genes predicted)*precision/recall].

AUCprecisionrecallgenes predictedgenes expected
Layers 2/30.7160%35%1831
Layer 40.8257%57%1111
Layer 50.7357%53%8086
Layer 60.5914%13%1516
Layer 6b0.7529%25%2428
Unpatterned0.6860%62%191185
Online Supplementary Table 4. Classifier statistics for patterning of astrocyte genes. Three statistics from each classifier are presented: Area Under the receiver operating characteristic Curve (AUC), proportion of the “genes predicted” that truly have elevated expression in the layer (precision), and the proportion of genes with elevated expression in the layer that are contained in the “genes predicted” (recall). The next column contains the number of genes predicted by each classifier out of 356 classifiable astrocyte-specific genes. The last column represents the number of genes expected to be preferentially expressed in this layer, and is calculated as [(genes predicted)*precision/recall].

Download
Online Supplementary Table 5. Mouse cortical lincRNA transcripts. Coordinates based on mouse genomic mapping (mm9).

Download
Online Supplementary Table 6. Known layer-enriched genes used to train classifiers. The Ensembl gene ID translations of layer-enriched genes are provided for genes annotated as being enriched in each layer. Additionally, the gene IDs of the genes not enriched in this layer are given. Note that some genes are enriched in multiple layers, and some genes are simply annotated as not being enriched in a specific layer (in the latter case, such genes are not included in any layer enrichments, but are considered patterned and are included as "not enriched" in the layer in which they are not enriched). The genes enriched in the layer that were expressed at sufficient level to compute relative expression levels are also provided along with their relative expression levels across samples, where the sum total of expression is 7 (corresponding to seven samples).

Download
Online Supplementary Table 7. Layer enrichment probabilities for neuron-specific genes. The default cutoff for enrichment probability, and the basis for the classifier statistics in Online Supplementary Table 3, was 0.5. Layer enrichment probabilities for neuron-specific genes were not calibrated.

Download
Online Supplementary Table 8. Layer enrichment probabilities for astrocyte-specific genes. The default cutoff for enrichment probability, and the basis for the classifier statistics in Online Supplementary Table 4, was 0.5. Layer enrichment probabilities for astrocyte-specific genes were not calibrated.

Download
Online Supplementary Table 9. Full list of significant functional enrichments amongst groups of layer-enriched genes. The ‘_cond’ suffix appended to some database names indicates that the enrichments were computed as a proportion of genes with that functional term that had some functional annotation in that database, instead of as a proportion of genes overall. Some databases were only used in this way, and do not contain the suffix to differentiate them (see Experimental Procedures for description of the sources of these and other databases).

Download
Online Supplementary Table 10. Full list of significant patterned & unpatterned functional enrichments. Positive values of log2(fold difference) correspond to significantly patterned functions, while negative values correspond to significantly unpatterned functions. See Experimental Procedures for description of the sources of these and other databases. Note that all rheumatoid arthritis (RA) genes in layers 2/3 genes are among the type 1 diabetes (T1D) genes. The lower fold enrichment for these RA-associated genes than for the T1D genes implies that this represents a more general association for layers 2/3 and T1D. Of course, the significance assigned to this particular enrichment (since the null hypothesis assumes independent expression of genes) is confounded by the concentration of T1D genes in the MHC regions of both mouse and human.

Download
Online Supplementary Table 11. Overlapping transcripts and normalized nucleotide substitution rates of cortical lincRNA loci. LincRNA transcripts and loci with associated normalized nucleotide substitution rates when aligned to human.

Download
Online Supplementary Table 12. Expression correlations between lincRNAs and neighboring protein-coding transcripts. LincRNAs are significantly more likely to be positively (or, less frequently, negatively) correlated in expression across samples with expression for their genomically neighboring protein-coding transcripts.

Download
Online Supplementary Table 13. Layer enrichment predictions for parentally-imprinted genes. Ensembl gene IDs, MGI gene IDs, gene symbols, gene names, and uncalibrated and calibrated layer enrichment probabilities for known genes identified as imprinted in mouse medial prefrontal cortex (Gregg et al.). Also includes the uncalibrated and calibrated layer enrichment probabilities for lincRNAs imprinted in mouse adult medial prefrontal cortex, adult preoptic area, or embryonic day 15 brain.

Download
Online Supplementary Table 14. Coordinates of candidate imprinted regions overlapping lincRNAs and lincRNAs with direct evidence for imprinting. Coordinates of imprinted regions that overlap lincRNAs along with the age/location in which significant parent-of-origin transcription was observed (mpfc = adult medial prefrontal cortex, poa = preoptic area of the adult hypothalamus, e15 = embryonic day 15 brain). Imprinted regions were assessed from published data (Gregg et al.) as described online. Also provided are lincRNA transcripts for which there was direct evidence of imprinting in mPFC, POA, or E15 brain.