Fig 1.
Overview of the data analysis pipeline.
A) Diagram of B-cell lymphopoiesis depicting the different B-cell subpopulations isolated from bone marrow and tonsils. B) Flow diagram highlighting the different steps for processing the exon array data. C) Illustration showing the concept of working with updated annotation and remapped probe sets for gene PI3, ENSG00000124102. Four defined exon probe sets cover the three exons. The most upstream probe set contains a probe that is not fully contained in the most current gene model of PI3 (highlighted in red). The remapped probe set combines all valid PI3 probes into a single probe set.
Fig 2.
Isolation of human B-cell subsets from bone marrow and tonsils.
A) Overlay of flow cytometry and array data on surface markers used for sorting of the bone marrow B-cell subsets. The contour diagrams show summary of events from the collected B-cell subsets in all samples and dots depict gene expression values (log2 intensities) in the individual cell samples. B) As in (A), but for tonsillar B-cell subsets and sorting markers. C) Expression profiles of selected genes, dots correspond to group means.
Fig 3.
Long noncoding RNA expression in human B-cell subpopulations.
A) Distribution of array-derived expression levels across all samples are shown for different gene biotype classes: Antisense, lincRNA, other lncRNA, mRNA, and pseudogene. B) Correlation of expression patterns between gene pairs located in close proximity on the genome. C) Hierarchical clustering of samples based on expression of protein-coding genes (top dendrogram) and all lncRNA classes (bottom dendrogram), respectively.
Fig 4.
Weighted gene co-expression network analysis of human B-cell subpopulation transcriptomes.
A) Cluster dendrogram showing genes grouped into distinct modules with height on the y-axis corresponding to co-expression distance between genes. B) Module expression summaries are shown with values of the components of the module eigengene (y-axis) versus microarray sample (x-axis).
Fig 5.
Connectivity of intramodular hub genes in three selected gene co-expression modules.
A) Highly connected genes in the brown module. B) Highly connected genes in the turquoise module. C) Highly connected genes in the yellow module. The node shapes indicate gene biotype, hexagon = antisense, octagon = lincRNA, circle = protein-coding, rounded rectangle = pseudogene, and rectangle = sense overlapping. The connectivity of a gene is encoded in node size with bigger nodes meaning higher connectivity. Edge transparency and width encode gene pair adjacencies, with thicker lines and lower transparency meaning higher similarity.
Table 1.
Module characteristics.
Fig 6.
Antisense RNAs in the brown module center.
A) Expression profiles and B) genomic organization of highly connected sense-antisense pairs from the brown module.
Fig 7.
Highly connected lincRNAs in the brown and yellow modules.
A) Genome browser plot for the highly connected lincRNA CTC-436k13.6 in the brown module. The EvoFold track shows position of a highly conserved RNA secondary structure that overlaps the exon-intron boundary. PhastCons scores show conservation calculated from a 100 species genome-wide multiple sequence alignment. DNaseI hypersensitive region tracks show data from i) CD20+ B-cells, ii) CD14+ monocytes, iii) CD34+ hematopoietic progenitor cells, and iv) Jurkat cell line. B) Expression profiles of RP11-132N15.3 and the nearby BCL6 gene. C) Visualization of the genomic region containing RP11-132N15.3 and BCL6.