Hiseq2500 rapid mode

Human dermal fibroblast (LONZA, CC-2509) and MEF (MEF_Ng-20D17, Riken BioResource Center) cell lines were used for evaluating the demultiplexing strategy. HDF and MEF cell lines were cultured in DMEM media with 10% fetal bovine serum and penicillin/streptomycin. For validation purpose, two MEF and two HDF plates were prepared for two rounds of C1 CAGE run. First assay was mixed with calcein green stained HDF and both red and green stained MEF. Second assay was mixed with calcein red stained MEF and both green and red stained HDF. After each staining, single-cell suspensions were prepared by trypsinization and washing. After count these cells by C-chip (NanoEnTek), each 1.5 × 10⁵ cells/ml were mixed in one tube. After cell capture and imaging, two rounds of C1 CAGE reaction were performed. Totally, 192 wells were indexed by index primer, dir #501-516/N701-N712, Supplementary Table 3). The final purified library was quality-controlled on a High-Sensitivity DNA Chip and quantified with the KAPA Quantification Kit. Nine pmol were sequenced and demultiplexed on single Illumina HiSeq 2500 Rapid mode (50 nt paired end). Reads were aligned to a combined mouse (mm10) and human (hg38) genome using the “CAGE processing” pipeline described above, and the fraction of reads uniquely mapping to each genome was calculated.

Mouse ESCs (B6G-2) cells purchased from Riken BioResource Center were maintained under feeder-free conditions in DMEM containing fetal bovine serum (GIBCO), l-glutamine, non-essential amino acids (GIBCO), 2-mercaptoethanol penicillin/streptomycin supplemented with leukemia inhibitory factor. Single-cell suspensions were prepared by accutase for 5 min at 37 °C. After count cells by C-chip (NanoEnTek) and adjustment of cell number, cell capture and live/dead staining were performed by Fluidigm C1 system. Followed by imaging, cDNA synthesis and library prep for C1 CAGE was performed as described above. The final purified library was quality-controlled on a high-sensitivity DNA Chip and quantified with the KAPA Quantification Kit. Totally, 9 pmol were sequenced and demultiplexed on single Illumina HiSeq 2500 Rapid mode (50 nt paired end). C1 STRT data were downloaded (E-MTAB-5482) and split into fastq files for each cell barcode. Read 1 of C1 CAGE data and C1 STRT data were aligned to the mouse genome (mm9) with STAR with parameters (--outFilterMultimapNmax 1 --outFilterScoreMinOverLread 0 --outFilterMatchNminOverLread 0) to select uniquely mapping reads. These reads were then annotated overlapping their 5 prime ends with FANTOM5 CAGE clusters. Strand invasion was calculated for each method using the R package CAGEr function findStrandInvaders with option linker = “GGG”.