idiffir: identifying differential intron retention from rna-seq
TRANSCRIPT
iDiffIR: Identifying differential intron retentionfrom RNA-seq
Michael Hamilton1, Anireddy SN Reddy2, Asa Ben-Hur1
Colorado State University1Dept of Computer Science, 2Dept of Biology
Hamilton et al iDiffIR
Introduction
Intron retention
Intron retention (IR) predominant AS form in plants
≈ 45% in plants, ≈ 9% in human
Important regulatory role in gene expression
Hamilton et al iDiffIR
Introduction
Differential IR
Lack of methods specifically for IRLittle consensus among published methods
Hamilton et al iDiffIR
Introduction
Differential IR
1
1Liu, R, Loraine, AE, Dickerson, JA (2014). Comparisons of computational methods for differential alternative
splicing detection using RNA-seq in plant systems. BMC Bioinformatics, 15:364.
Hamilton et al iDiffIR
Results
Summary of datasets
source condition species length reads (mills)met1 untreated col-0
Arabidopsis 10156.8
untreated met1 47.4
PTB 6 libraries
Arabidopsis 80/101 ≈ 60(3 conds X 2 reps)
hnRNP
U
hnRNP U CTL∗human 75
28.2hnRNP U KD∗ 28.6
Hamilton et al iDiffIR
Results
Significant differential IR
Dataset iDiffIR MATS MISO(10)
met1 349 20 33ptb-mi12 r. 1 324 45 127ptb-mi12 r. 2 350 0 34ptb-mi12 529 - -ptb-oe1 r. 1 435 0 73ptb-oe1 r. 2 338 0 46ptb-oe1 583 - -hnRNP U 224 9 38
Hamilton et al iDiffIR
Results
met1 mutant
Dataset iDiffIR MATS MISO(10)
met1 349 20 332
ptb-mi12 r. 1 324 45 127ptb-mi12 r. 2 350 0 34ptb-mi12 529 - -ptb-oe1 r. 1 435 0 73ptb-oe1 r. 2 338 0 46ptb-oe1 583 - -hnRNP U 224 9 38
Overrepresented GO categories for abiotic and biotic stress
met1 mutant resistant to bacterial pathogens
Exhibits up/down regulation of pathogen responsive genes
2Dowen, RH, et al. (2012). Widespread dynamic DNA methylation inresponse to biotic stress. Proc. Natl. Acad. Sci. U.S.A., 109, 32:E2183-91.
Hamilton et al iDiffIR
Results
PTB KD/OE lines
Dataset iDiffIR MATS MISO(10)
met1 349 20 33ptb-mi12 r. 1 324 45 1273
ptb-mi12 r. 2 350 0 34ptb-mi12 529 - -ptb-oe1 r. 1 435 0 73ptb-oe1 r. 2 338 0 46ptb-oe1 583 - -hnRNP U 224 9 38
529/583 differential IR events in KD and OE lines
GO analysis found nucleotide binding top molecular functionterm (FDR adj p-value of 1.3 × 10−4)
3Ruhl, el al. (2012). Polypyrimidine tract binding protein homologs fromArabidopsis are key regulators of alternative splicing with implications infundamental developmental processes. Plant Cell, 24, 11:4360-75.
Hamilton et al iDiffIR
Results
PTB validation
Gene Model for AT1G22740
100101102
MI12 Rep 1
100101102
MI12 Rep 2
100101102
Wildtype Rep 1
8049200 8049400 8049600 8049800 8050000 8050200 8050400 8050600100101102
Wildtype Rep 2
Hamilton et al iDiffIR
Results
hnRNP U KD
Dataset iDiffIR MATS MISO(10)
met1 349 20 33ptb-mi12 r. 1 324 45 127ptb-mi12 r. 2 350 0 34ptb-mi12 529 - -ptb-oe1 r. 1 435 0 73ptb-oe1 r. 2 338 0 46ptb-oe1 583 - -hnRNP U 224 9 384
224 differential IR events
Terms associated with RNA-binding, translation, andribosome occupancy
4Xiao, R, et al. (2012). Nuclear matrix factor hnRNP U/SAF-A exerts aglobal control of alternative splicing by regulating U2 snRNP maturation. Mol.Cell, 45, 5:656-68.
Hamilton et al iDiffIR
Results
hnRNP U KD
Gene Model for ENSG00000153187
100
101
102
103hnRNP Rep 1
245016000245018000245020000245022000100
101
102
103Control Rep 1
Hamilton et al iDiffIR
ARAPORT
ARAPORT annotations
Annotations
Hamilton et al iDiffIR
ARAPORT
iDiffIR results on ARAPORT
Results
Hamilton et al iDiffIR
ARAPORT
Multi condition iDiffIR results
Gene Model for AT3G01770
100
101
SR45_KD Rep 1
100
101
SR45_KD Rep 2
100
101
Wildtype Rep 1
275500276000276500277000277500278000278500100
101
Wildtype Rep 2
Gene Model for AT3G01770
100
101
102 Triple Rep 1
100
101
102 Triple Rep 2
100
101
102 Wildtype Rep 1
275500276000276500277000277500278000278500
100
101
102 Wildtype Rep 2
Hamilton et al iDiffIR
ARAPORT
iDiffIR as an ARAPORT app
No need to install (Dependencies!)
Ease of use
Data management–iPlant access
ThaleMine access
Hamilton et al iDiffIR
ARAPORT
iDiffIR as an ARAPORT app
Hamilton et al iDiffIR
Summary
Summary
Method specific for identifying differential IR
Robust to expression, replicates, and library size
Detects biologically-significant differential IR
Extensively tested on several organisms
ARAPORT integration: annotations, displaying analyses,science app
Availability: https://bitbucket.org/comp_bio/idiffir
Hamilton et al iDiffIR
Summary
Acknowledgements
Asa Ben-Hur ASN Reddy
Funding: USDA/DOE Plant Feedstocks Genomics for Bioenergy
SpliceGrapher: Mark Rogers
Hamilton et al iDiffIR
Summary
Validation in rice: annotated IR
Hamilton et al iDiffIR
Method
Quantifying IR
Avg Read Depth
Avg Read Depth
We use a log-fold change statistic of average read depth, adjusted for library size, differential gene expression, and exonic read depth variability.
Hamilton et al iDiffIR
Method
Sources of bias
Library size
DE Genes
Library size
DE Genes
Hamilton et al iDiffIR
Method
Measuring variance
E1 E2 E3 E4
Hamilton et al iDiffIR
Method
Detecting events across expression levels
a
4 2 0 2 4 6 8 10 128
6
4
2
0
2
4
6
8a=22
a=23
a=24
a=25
a=26
Expression Levello
g F
C
Hamilton et al iDiffIR