SQANTI: Classification, Curation and Quantification of a PacBio transcriptome

Size: px

Start display at page:

Download "SQANTI: Classification, Curation and Quantification of a PacBio transcriptome"

Jeffery Reeves
6 years ago
Views:

1 SQANTI: Classification, Curation and Quantification of a transcriptome Manuel Tardaguila, PhD Ana Conesa Lab

2 in Oligodendrogenesis Neural Stem Cells (neurospheres) Oligodendrocyte Progenitor Cells Immature Oligodendrocytes Premyelinating mature Oligodendrocytes Myelinated mature Oligodendrocytes cdna cdna Illumina Short reads RoIs ToFU pipeline RefSeq Mapping Indels ToFU transcripts Curated Transcriptome QC features + Machine Learning Corrected Transcriptome SQANTI Global Transcriptome Expressed Transcriptomes Tardaguila et al SQANTI: extensive characterization of long read transcript sequences for quality control in full-length transcriptome identification and quantification Pre-print BioRxiv (2017)

3 1. CLASSIFICATION OF PACBIO TRANSCRIPTS

4 Splice-based classification (I): Known Transcripts Full Splice Match (FSM) Reference Donor Acceptor Splice Junction Reference n= 16,104 Incomplete Splice Match (ISM) Reference Reference

5 Splice-based classification (II): Novel Transcripts from known genes Novel In Catalog (NIC) Novel SJs with known Donors and acceptors Reference New combination of Known SJs Reference n= 16,104 Retained Intron Novel Not in Catalog (NNC) Novel SJs from novel Donors and/or Acceptors Reference

6 Splice-based classification (III): Novel Transcripts from novel genes Intergenic Gene A Gene B n= 16,104 Genic Intron Exon A Intron A-B Exon B Fusion Transcripts Gene A Gene B

7 Splice-based classification (IV): Antisense and Genic Genomic transcripts Genic Genomic Exon A Intron A-B Exon B n= 16,104 Antisense Gene A Antisense Ref 1 Ref 2

8 CLASSIFICATION SUMMARY Splice based classification allows to separate known transcripts (FSM and ISM) from novel transcripts arising from novel splice junctions or new combinations of already known splice junctions (NIC and NNC) The predominant categories in our transcriptome are FSM,ISM, NIC and NNC and they are mostly coding (oligodt library preparation)

9 2. CURATION OF PACBIO TRANSCRIPTS

10 40% novel isoforms in mouse... Are all of them real?

11 output needs curation NNC transcripts characterized by novel Donors and/or Acceptors of splicing concentrate traits of low quality transcripts

12 Using Short reads and Long reads to mine QC features Corrected Transcriptome

13 Features selected for Random Forest Classification bite MinCov Min_sample_cov FL ratioexp as.factor(testing$bite) FALSE TRUE NEG POS NEG POS test set test set geneexp 5e+01 5e+02 5e+03 5e+04 5e NEG POS NEG POS test set test set isoexp MinCovPos NEG POS NEG POS NEG POS test set test set test set nindels NEG POS test set sdcov FSM_class testing[, i] A B C NEG POS NEG POS test set test set length exons nindelsjunc NEG POS NEG POS NEG POS NEG POS test set test set test set test set

PCR validation in an independent set of transcripts FSM NIC NNC Fusion oligodt R.H. Temp ( C) Nolc1 + - 42 - + 42 - + 50 PB.

oligodt R.H. Temp ( C) Trp53inp2 + - 42 - + 42 - + 50 PB.4207.

14 PCR validation in an independent set of transcripts FSM NIC NNC Fusion oligodt R.H. Temp ( C) Nolc PB NNC isoform oligo (dt) PCR Positive Negative Total 23 (3nc) (3 nc) 12 (8 nc) pb 500 pb oligodt R.H. Temp ( C) Trp53inp PB NNC isoform 729 pb 1000 pb Positive 5 (3nc) pb * 895 pb R.H. PCR Negative (3nc) 0 oligodt R.H. Temp ( C) Pgam PB NNC isoform Total pb * 463 pb

15 Splice Junction independent curation: Intrapriming features % of intrapriming Filtered out in each category n=35 n=81 n=135 n=17 n=108 n=17 n=210 FSM ISM NIC NNC Genic Genomic Antisense Fusion Intergenic Genic Intron Non Coding Coding oligodt can prime outside polya regions in A rich regions inside transcripts We looked for transcripts showing >= 80% Adenines in the 20 nts downstream (DNA) the end of thetranscript If this transcripts lack a consensus poly Adenilation site we filter them out 605 transcripts filtered out

16 n= 12,908 SQANTI filter eliminates transcript with bad QC features

17 SQANTI filter works across different datasets: Maize Before SQANTI AFTER SQANTI

18 SQANTI filter works across different datasets: MCF-7 Before SQANTI AFTER SQANTI

19 CURATION SUMMARY output needs curation, specially in NNC novel acceptor/ donnor transcripts SQANTI mines features based in splice junctions, expression and structure to feed a RF classifier that succeeds in filtering out transcripts that fail to validate in an independent PCR set SQANTI works across different datasets

20 3. QUANTIFICATION OF PACBIO TRANSCRIPTOME

21 Quantification: curated vs RefSeq: Transcripts High Exp n=6,357 Medium Exp n=5,677 Low Exp n=422 21,934 8,782 3, Transcript expresion (log(tpm)) OLP replicate 1 R 2 = 0.99 R 2 = 0.89 R 2 = 0.39 OLP replicate 2 RefSeq High Exp n=7,440 Found by both Medium Exp n=7,440 Exclusive Low Exp n=8,379 OLP replicate 1 R 2 = 0.99 R 2 = 0.88 R 2 = 0.3 OLP replicate 2 Found by both RefSeq Exclusive finds a set of highly expressed transcripts and reproducible transcripts in common with RefSeq RefSeq detects lots of lowly expressed, difficult to reproduce transcripts. However it also captures exclusively 30% of highly expressed ones ( 18%).

22 Most of the exclusive transcripts are new combinations of already known Splice Junction 60 3,674 exclusive transcripts % of novel transcripts i Exclusive ENSEMBL NIC: new combination of known SJ NIC: monoexon by IR NIC: novel SJ of known D/A NNC antisense fusion genic genomic genic intron intergenic

23 Quantification: curated vs RefSeq: Genes High Exp n=2,729 Medium Exp n=4,181 Low Exp n= , , Transcript expresion (log(tpm)) OLP replicate 1 R 2 = 0.99 R 2 = 0.99 R 2 = 0.92 OLP replicate 2 RefSeq High Exp n=3,330 Found by both Medium Exp n=6,589 Exclusive Low Exp n=3,315 OLP replicate 1 R 2 = 0.99 R 2 = 0.98 R 2 = 0.93 OLP replicate 2 Found by both RefSeq Exclusive RefSeq detects lots of lowly expressed genes. However it captures an important fraction of highly expressed ones 19% ( 1.5%)

24 Imposing a higher EXP threshold highlights advantages and disadvantages of quantification 5,101 7,425 2, Transcript expresion (log(tpm)) OLP replicate 1 OLP replicate 1 R 2 = 0.99 R 2 = 0.89 R 2 = OLP replicate 2 RefSeq R 2 = 0.99 R 2 = 0.87 R 2 = 0.22 OLP replicate 2 High Exp n=6,350 High Exp n=7,339 Found by both Medium Exp n=3,167 Medium Exp n=5,109 Found by both Low Exp n=19 Exclusive Low Exp n=78 RefSeq Exclusive % of novel transcripts i Exclusive ,111 exclusive transcripts ENSEMBL NIC: new combination of known SJ NIC: monoexon by IR NIC: novel SJ of known D/A NNC antisense fusion genic genomic genic intron intergenic ,110 5,838 Transcript expresion (log(tpm)) OLP replicate 1 OLP replicate 1 R 2 = 0.99 R 2 = 0.98 R 2 = 0.98 OLP replicate 2 RefSeq R 2 = 0.99 R 2 = 0.98 R 2 = 0.95 High Exp n=2,698 High Exp n=3,268 Found by both Medium Exp n=3,725 Medium Exp n=4,488 Low Exp n=7 Exclusive Low Exp n=186 % of total Genes % of total Genes Isoforms per gene >= 6 RefSeq Isoforms per gene >= 6 OLP replicate 2 Found by both RefSeq Exclusive 0

25 Analysis of Most Expressed Transcript (MET) reveal unaccounted 3 end variability that is captured by Dhrs7b - Dehydrogenase / reductase (SDR family) member 7B ExR GlR Read Coverage Mean Transcript expresion (TPM) NM_ / PB NM_ PB NM_ PB NM_ / PB XM_ pb 1000 pb 1500 pb 1000 pb XM_ pb 1000 pb Lowest SJ coverage by short reads ns Same MET *** Different MET Lowest mean coverage of an exon ns Same MET *** Different MET ExR GlR Distance to reference TTS (nts) Downstream Upstream *** Same MET Different MET

26 CUANTIFICATION SUMMARY captures a robust fraction of transcripts and genes being expressed and at the same time allow for novel discovery. RefSeq captures more transcripts and genes, many lowly expressed and hardly reproducible ones However, TOFU fails to capture many highly expressed genes detected by RefSeq. transcriptome reveals unaccounted for 3 end variability in known transcripts that hamper RefSeq quantification. Tardaguila et al SQANTI: extensive characterization of long read transcript sequences for quality control in full-length transcriptome identification and quantification Pre-print BioRxiv (2017)

27 4. Functional outcome of alternative splicing: Transcript2GO

analytical possibilities Transcript2GO: assessing the

28 T2GO combines SQANTI classification with Genomic, transcript and protein annotation to maximize analytical possibilities Transcript2GO: assessing the functional outcome of alternative splicing manuscript in preparation

Diferential splicing linked to the appearance of

Nuclear Fraction Cytosolic Fraction M NPC1 NPC2 OPC1

GAPDH ORF collapse H3-Ac *** Transcript2GO: assessing

29 Diferential splicing linked to the appearance of sequence motifs on a transcriptome wide scale ns INPUT Nuclear Fraction Cytosolic Fraction M NPC1 NPC2 OPC1 OPC2 M NPC1 NPC2 OPC1 OPC2 M NPC1 NPC2 OPC1OPC2 *** GAPDH ORF collapse H3-Ac *** Transcript2GO: assessing the functional outcome of alternative splicing manuscript in preparation

30 Thanks!

RNA standards v May

RNA standards v May Standards, Guidelines and Best Practices for RNA-Seq: 2010/2011 I. Introduction: Sequence based assays of transcriptomes (RNA-seq) are in wide use because of their favorable properties for quantification,