Ginkgo Interactive analysis and quality assessment of single-cell CNV data

Transcription

1 Ginkgo Interactive analysis and quality assessment of single-cell CNV Robert Aboukhalil, Tyler Garvin, Jude Kendall, Timour Baslan, Gurinder S. Atwal, Jim Hicks, Michael Wigler, Michael C. Schatz

2 Outline Introduction Comparison of WGA methods Ginkgo

3 Outline Introduction Comparison of WGA methods Ginkgo

4 Single-cell sequencing Circulating tumor cells Neuronal mosaicism Clonal evolution in tumors Recombination/ crossover in germ cells

5 Single-cell vs. bulk sequencing Brian Owens, Nature News 2012

6 Single-cell vs. bulk sequencing Brian Owens, Nature News 2012

7 Copy-number variant analysis Low coverage allows us to study copy-number variants <1X$coverage,$o-en$<0.1X$

8 Copy-number variant analysis NNNNNN" Low coverage allows us to study copy-number variants <1X$coverage,$o-en$<0.1X$

9 Copy-number variant analysis NNNNNN" Divide genome into bins with ~ reads / bin Baslan et al., Nature Protocols 2012

10 Copy-number variant analysis NNNNNN"

11 Copy-number variant analysis NNNNNN"

12 Copy-number variant analysis NNNNNN" Circular Binary Segmentation (CBS) to reduce noise in data

13 Copy-number variant analysis We can estimate integer copy-number states by scaling the profile and minimizing the sum of squares error

14 Outline Introduction Comparison of WGA methods Ginkgo

15 Copy-number variant analysis

16 Analysis Pipeline Bin reads Map reads GC bias correction.fastq$ Remove duplicates BAM to BED.BED$ Segment bins and call integer copy-number Galaxy Interactive visualization Ginkgo

17 Ginkgo Demo

18 Sample dataset Triple-negative (ER, PR, HER2 ) ductal carcinoma Navin et al., Nature 2011

19 QB.CSHL.EDU/GINKGO

20 QB.CSHL.EDU/GINKGO

21 QB.CSHL.EDU/GINKGO

22 QB.CSHL.EDU/GINKGO

23 Diploids from both primary and metastasis cluster QB.CSHL.EDU/GINKGO

24 Diploids from both primary and metastasis cluster QB.CSHL.EDU/GINKGO

25 Non-diploids from primary and metastasis segregate QB.CSHL.EDU/GINKGO

26 Non-diploids from primary and metastasis segregate QB.CSHL.EDU/GINKGO

27 Non-diploids from primary and metastasis segregate QB.CSHL.EDU/GINKGO

28 Non-diploids from primary and metastasis segregate QB.CSHL.EDU/GINKGO

29 QB.CSHL.EDU/GINKGO

30 QB.CSHL.EDU/GINKGO

31 QB.CSHL.EDU/GINKGO

32 QB.CSHL.EDU/GINKGO

33 QB.CSHL.EDU/GINKGO

34 QB.CSHL.EDU/GINKGO Although primary and metastasis cells are similar, they are slightly different from each other

35 QB.CSHL.EDU/GINKGO Heatmap also tells us there are 3 sub populations: 1 diploid, 1 primary, 1 metastasis

36 QB.CSHL.EDU/GINKGO Can click on cells to get indepth view

37 QB.CSHL.EDU/GINKGO

38 QB.CSHL.EDU/GINKGO Interactive profile viewer allows you to zoom into a region of interest

39 QB.CSHL.EDU/GINKGO

40 QB.CSHL.EDU/GINKGO We automatically upload amplification and deletion tracks to the UCSC browser

41 QB.CSHL.EDU/GINKGO We can also annotate the CNV profile with genes of interest

42 QB.CSHL.EDU/GINKGO

43 QB.CSHL.EDU/GINKGO

44 QB.CSHL.EDU/GINKGO

45 QB.CSHL.EDU/GINKGO

46 QB.CSHL.EDU/GINKGO

47 QB.CSHL.EDU/GINKGO

48 QB.CSHL.EDU/GINKGO

49 QB.CSHL.EDU/GINKGO Static CNV profile

50 QB.CSHL.EDU/GINKGO

51 Original read count data before analysis QC metrics QB.CSHL.EDU/GINKGO

52 QB.CSHL.EDU/GINKGO GC bias, before and after correction

53 QB.CSHL.EDU/GINKGO How we infer integer copy-number state using sum of squares error

54 Outline Introduction Comparison of WGA methods Ginkgo

55 Whole Genome Amplification (WGA) methods DOP-PCR (Degenerate Oligonucleotide Primed PCR) MDA (Multiple Displacement Amplification) MALBAC (Multiple Annealing and Looping Based Amplification Cycles) Paul Blainey, FEMS Microbiol Rev. 2013

56 Comparison of WGA methods Paper WGA Method Tissue Navin et al., 2011 DOP-PCR Breast (T10) Navin et al., 2011 DOP-PCR Breast (T16P/M) McConnnell et al., 2013 DOP-PCR Neuron Lu et al., 2012 MALBAC Sperm Ni et al., 2013 MALBAC Lung Hou et al., 2013 MALBAC Oocyte Kirkness et al., 2013 MDA Sperm Wang et al., 2012 MDA Sperm Evrony et al., 2012 MDA Neuron Explore the effects of WGA method on data quality: 1) GC bias 2) Coverage dispersion Garvin and Aboukhalil et al., Nature Methods, 2015

57 GC Bias Garvin and Aboukhalil et al., Nature Methods, 2015

58 Coverage Dispersion DOP-PCR MALBAC MDA Garvin and Aboukhalil et al., Nature Methods, 2015

59 Coverage Dispersion DOP-PCR MALBAC MDA Garvin and Aboukhalil et al., Nature Methods, 2015

60 Summary Ginkgo is a platform for single-cell CNV analysis and visualization For copy-number analysis, we recommend DOP-PCR Check out Ginkgo and give us feedback qb.cshl.edu/ginkgo Garvin and Aboukhalil et al., Nature Methods, 2015 > 1K USERS 21,500 PAGEVIEWS$ 10 MIN TIME SPENT Nov 3, San Francisco

61 Thanks Ginkgo Team Tyler Garvin Jude Kendall Timour Baslan Jim Hicks Gurinder S. Atwal Michael Wigler Michael C. Schatz qb.cshl.edu/ginkgo