 |
 |
GEO help: Mouse over screen elements for information. |
|
| Status |
Public on Feb 02, 2017 |
| Title |
SPAR-Seq of multi-exon regions of 50 genes in CGR8, Arid3b knockdown (T0019), replicate 1 |
| Sample type |
SRA |
| |
|
| Source name |
Embryonic stem cells
|
| Organism |
Mus musculus |
| Characteristics |
cell line: CGR8
|
| Treatment protocol |
Twenty-four hours prior to transfection, CGR8 and N2A cells were seeded in 96-well plates, 3000 and 5000 cells per well, respectively. Cells were transfected with the SMARTpool siRNAs at 50 nM final concentration using DharmaFECT1 reagent (Dharmacon), as recommended by the manufacturer.
|
| Growth protocol |
CGR8 mouse embryonic stem cells (ESCs) were cultured as described previously on gelatin-coated plates (Gabut et al., Cell (2011), Han et al., Nature (2013)). Mouse neuroblastoma (N2A) cells were grown in DMEM supplemented with 10% FBS, sodium pyruvate, MEM non-essential amino acids, and penicillin/streptomycin. All cell lines were maintained at 37°C with 5% CO2.
|
| Extracted molecule |
total RNA |
| Extraction protocol |
Forty-eight hours post-transfection, total RNA were purified from cultured cells using RNeasy Plus 96 Kit (Qiagen), as per the manufacturer’s instructions.
For each sample, a multiplex RT-PCR assay was first applied to simultaneously amplify 50 AS and/or expression events in a single reaction. The multiplex RT-PCR reaction was carried out in 96-well plates using the OneStep RT-PCR kit (Qiagen) as recommended by the manufacturer, with the following changes: reactions were performed in a volume of 20 uL with 2 uL of the purified total RNA as input, and a mixture of 50 pairs of primers was added to each reaction with a final concentration of 0.25 uM for each individual forward and reverse primer. Four identical Veriti 96-well Thermal Cyclers (Applied Biosystems) were used with the following conditions: 50°C for 30 minutes, 95°C for 15 minute, 30 cycles of 94°C for 40 seconds, 58°C for 1 minute (slow ramp rate), 72°C for 3 minutes, and a final extension step at 72°C for 10 minutes. Two sets of barcodes were incorporated into forward and reverse primers, respectively, after the universal adaptors and added to the amplicons in the second PCR reaction to allow demultiplexing all reads that were derived from a particular treatment. Primers sequences were: forward primer, AATGATACGGCGACCACCGAGATCTACACNNNNNNNNACACTCTTTCCCTACACGACGCTCTTCCGATCT (N denotes i5 barcode); reverse primer, CAAGCAGAAGACGGCATACGAGATNNNNNNNNGTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT (N denotes i7 primer). The second PCR was performed using Phusion High-Fidelity DNA Polymerase (Thermo Scientific), as per the manufacturer’s instructions. For each 20 uL of reaction, 1 uL of the multiplex RT-PCR reaction products was used as templates. The thermal cycling conditions were as follows: 98°C for 30 seconds, 15 cycles of 98°C for 10 seconds, 65°C for 30 seconds, 72°C for 30 seconds, and a final extension step at 72°C for 5 minutes. The resulting libraries were pooled and sequenced.
|
| |
|
| Library strategy |
OTHER |
| Library source |
transcriptomic |
| Library selection |
other |
| Instrument model |
Illumina HiSeq 2500 |
| |
|
| Description |
Forward (i5) barcode: TAGATCGC, reverse (i7) barcode: ATTAGGCC
KnockdownAnnotation.tab
ScreenJuncUp.20-77.fa
ScreenJuncDn.20-77.fa
ScreenEvents_junctions.tab
PSI.SSMD.CGR8.tab
Reads_perEvent_CGR8.tab
Expression.CGR8.treatments.tab
|
| Data processing |
CASAVA 1.8 was used for base calling.
Fastq files contain demultiplexed reads which were assigned to the appropriate treatment by matching to expected forward and reverse barcodes (sequenced in dedicated reads; not supplied).
Forward and reverse event reads were mapped to custom junction libraries representing all expected splice variants (supplied as processed files), using bowtie with settings --best -v 3 -k 1 --trim3 26 --trim5 20. Trimming was done to remove lower-quality, uninformative ends and because forward reads had a missing base call at position 11 due to a power outage. Junction libraries were first constructed from NCBIm37/mm9 gene annotations, and then refined based on the results of de novo mapping of the reads from one full Illumina lane from each cell line using tophat (default settings with -i 40 and providing Ensembl transcript annotations for NCBIm37), after quality trimming (minimum MAPQ of 33) from the 3'-end with the fastx toolkit. Mapped reads were counted per junction.
Alternative splicing analysis: Percent-spliced-in (PSI) values were calculated for each alternative exon, or part thereof in the case of alternative 5'-/3'-splice sites ('event'), from read tallies as follows: PSI was calculated independently from forward and reverse reads as the percentage of reads supporting inclusion divided by the total number of reads for the event. For alternative 5'/3'-ss events that were part of an alternative exon (Foxm1.A2, Mta1.A1, Uspl1.A3), the PSI was instead calculated with reference to the total reads supporting inclusion of that exon in order to assess independent regulation of alternative splice site usage. Only the read direction in which all reads unambiguously supported either inclusion or exclusion of an event was used, and the average if that applied for both. Exploratory analysis revealed that PSI values showed low-level batch effects per 96-well plate which we moderated by subtracting a weighted plate median in which negative controls (siNT and mock treatment) were given 20x more weight than other samples. To derive variances, and because PSI values are not nearly normally distributed but roughly follow a beta distribution, we elected to fit a beta distribution to each pair of replicate treatments, as well as to all negative controls from both replicates, using maximum-likelihood fitting as implemented in the fitdistr() function from the R package MASS {Venables & Ripley, Modern Applied Statistics with S (2002)}. Iterative optimization of shape parameters was initiated with settings x=PSI, shape1=1, shape2=1, method=”L-BFGS-B”, lower=0.01, and upper=mean number of reads supporting each PSI value. A modified Strictly Standardized Mean Difference (SSMD {Zhang, Genomics (2007)}) was then calculated based on the formula: SSMD = (μt – μc)/SQRT(vart + varc) where μt and μc are the means (corresponding to the PSI), and vart and varc the variances of the beta distributions fitted to the treatment replicates and negative controls, respectively. Events for which not all reads from at least one direction were informative (Mff.A2, Tead1.A2) were excluded from further analysis, as were events in individual treatments with less than 20 reads in one or both replicates (~9% of all event x treatment combinations). Additionally, the following events were removed from further splicing analysis: expression only events or constitutive exons, Fgf4, Gapdh, Srpk2, Srrm4; events with consistently low read counts: Atg13.A1, H2afy.A2, Tcf7l1 (N2A only), Dnmt3b (N2A only); or because measured inclusion was biased by differential length of isoforms:, Fgfr1, Sall4.
Expression analysis: Read counts from all splice variants of each gene were used to estimate relative mRNA expression levels, reads per million reads (RPMs). Differential expression analysis based on raw read counts was performed using the generalized linear models workflow from the R package edgeR. Plate, position (edge/interior) and treatment were used as design factors (where all treatments of each type of control were treated as replicates.) Models fitted using estimateDisp() to estimate the common, trended, and tag-wise dispersion for the CGR8 and N2A data separately. Subsequently, differences attributable to treatment, plate, or position contrasts were extracted with glmLRT() and represented as log2-fold changes with associated FDR. Fold-changes for treatments are relative to the siNT and mock-treatment controls.
Genome_build: NCBIm37/mm9
Supplementary_files_format_and_content: Processed files are either tab-delimited tables (.tab) or FASTA files (.fa). KnockdownAnnotation.tab contains treatment annotations and well/plate positions. ScreenJuncUp.20-77.tab and ScreenJuncDn.20-77.tab contain the custom junction upstream and downstram junction libraries, respectively, and ScreenEvent_junctions.tab contains the information which junctions were used to calculate PSI for each event. PSI.SSMD.CGR8.tab and PSI.SSMD.N2A.tab contain raw PSI of each replicate and final SSMD. Reads_perEvent_CGR8.tab and Reads_perEvent_N2A.tab contain raw read counts on which expression analyis was based, and Expression.CGR8.treatments.tab and Expression.N2A.treatments.tab contain modeled relative abundance and fold-change with associated FDR for each treatment type.
|
| |
|
| Submission date |
Apr 12, 2016 |
| Last update date |
May 15, 2019 |
| Contact name |
Ulrich Braunschweig |
| Organization name |
University of Toronto
|
| Department |
Donnelly Centre
|
| Lab |
Benjamin J. Blencowe
|
| Street address |
160 College Street
|
| City |
Toronto |
| State/province |
Ontario |
| ZIP/Postal code |
M5S 3E1 |
| Country |
Canada |
| |
|
| Platform ID |
GPL17021 |
| Series (2) |
| GSE80196 |
Multilayered control of alternative splicing regulatory networks by transcription factors (SPAR-Seq) |
| GSE80205 |
Multilayered control of alternative splicing regulatory networks by transcription factors |
|
| Relations |
| BioSample |
SAMN04693156 |
| SRA |
SRX1694244 |
| Supplementary data files not provided |
SRA Run Selector |
| Raw data are available in SRA |
| Processed data are available on Series record |
|
|
|
|
 |
