Content area
Full text
min-intron-length was set up to 30 nucleotides according to the intron length described by Moss et al. [84]. For the mouse and human datasets, raw data were downloaded from the public databases: human pancreas (four samples from Fagerberg et al. [37] and one acinar cell-enriched sample from Morán et al. [9]: the E-MTAB-1294 dataset), human islets (three samples from Nica et al. [46] and four samples from Morán et al. [9]), murine pancreas (two samples from Holmstrom et al. [36]), and murine islets (three samples from Morán et al. [9]). The alpha and beta cell RNA-seq datasets were obtained from Benner et al. [12] for mouse (two samples of each) and from Blodgett et al. [48] for human (six samples each) [48]. Reads were mapped into the mouse GRCm38 and the human GRCh37 genomes (Ensembl genome version 75) using default options. Gene expression was measured from the mapped reads by using HT-seq-count [85]. PCA, using princomp function of R, was calculated for the whole dataset using the variance stabilization transformation values obtained by DESeq R package from the gene expression values [86]. For differential expression analysis, we used the R package DESeq2 [87]. DESeq2 employs shrinkage estimation for dispersions and fold change; it uses Wald test for significance with posterior adjustment of P values using the procedure of Benjamini and Hochberg (giving adjusted P values). Genes differentially expressed were selected with an adjusted P?<?0.05 and a fold change?>?4. For the comparison of the acinar, ductal and endocrine cell data, in-silico endocrine datasets were simulated by combining alpha, beta and delta RNA-seq data. This approach was taken to decrease the number of pairwise comparisons and could be used since the sequencing deepness of each library was in a similar range (below of a factor 2). Endocrine dataset #1 was obtained by combining the mapped reads of alpha1 (47?×?106 reads), beta1 (48?×?106 reads), and delta1 (31?×?106 reads) datasets; endocrine dataset #2 is a mix of alpha2 (46?×?106 reads), beta2 (29?×?106 reads), and delta2 (30 106 reads); endocrine dataset #3 is the combination of alpha3 (59?×?106 reads), beta3 (62?×?106 reads), and delta3 (59?×?106 reads) datasets. The RNA-seq raw data have been deposited on ENA under the accession number PRJEB10140.
Comparison of the human, murine, and zebrafish pancreatic transcriptomes