Keller_ATAC_HA_072115_v2.pptx

Download Report

Transcript Keller_ATAC_HA_072115_v2.pptx

ATAC-seq on hematopoietic
progenitors
Cheryl Keller
The Pennsylvania State University
07-21-15
PSU
Ross Hardison
Belinda Giardine
Maria Long
NHGRI
David Bodine
Elisabeth Heuston
ATAC-seq – basic protocol
Wet lab
1.
2.
3.
4.
5.
6.
Collect and lyse cells
Transposition reaction and purification
PCR amplify for 5 cycles
Perform a qPCR side reaction to reduce bias from size and
GC content.
Perform additional cycles of amplification (as per Step 4)
Broad size selection (Hardison lab modification)
Data processing
1.
2.
3.
4.
5.
6.
7.
Map reads to mm9 genome using Bowtie
Use bedtools to convert bam to bed
Analyze mapped reads using Terry Furey’s F-seq
Obtain the F-seq peak calls (narrowPeak)
Filter to remove blacklisted regions
Mainly examine the signal tracks (wig to bigWig from
F-seq) in the following analyses
Peak analyses limited to top 100K peaks.
7/12/2016
2
qPCR side reaction to reduce bias from size and GC content
1. PCR (50 ml) amplify for 5 cycles
2. Remove 5 ml of amplified product for a qPCR side reaction
3. Determine the additional cycles needed for the remaining 45 ml PCR reaction as follows:
a. Plot linear Rn vs Cycle
b. Calculate the # of cycles that corresponds to ¼ of max fluorescent intensity
7/12/2016
3
ATAC-seq reveals regulatory landscape: Constitutive and
progenitor specific
7/12/2016
4
ATAC-seq reveals regulatory landscape: ERY vs MEG specific
7/12/2016
5
What factors influence quality of ATAC-seq data?
• Cell type?
– Bone marrow derived hematopoietic progenitors
• Cell collection procedures?
– FACS sorted into DMEM w/ 10% FBS
• Starting number of cells?
Performed a mini titration
with G1E-ER4+E2 cells
– Protocol calls for 50,000
• Amount of transposase?
– Protocol calls for 2 mL Tn
• Extent of PCR amplification?
ID
808
809
810
811
812
813
814
Tn (ul)
1
2
3
2
2
2
2
Cell #
50,000
50,000
50,000
100,000
150,000
50,000
50,000
PCR cycles
+n
+n
+n
+n
+n
+(n+3)
+(n+6)
n = Number of additional cycles according to
PCR side reaction following the original protocol
7/12/2016
6
Titration datasets - G1E-ER4+E2
7/12/2016
7
mm9
50 kb
Scale
chr7:
110,950,000
111,000,000
Custom annotation of genes and regulatory regions for Hbb locus from the literature
111,050,000
knCRMs
0.1 _
UCSC Genes (RefSeq, GenBank, tRNAs & Comparative Genomics)
Beta-s
Hbb-bh1
Olfr66
Gm5736
Hbb-y
Hbb-bh1
808: ATAC-seq on G1E-ER4+E2 cells Tn1 50,000 cells PCRn
0_
0.1 _
809: ATAC-seq on G1E-ER4+E2 cells Tn2 50,000 cells PCRn
Olfr68
Hbb-b1
Olfr67
Olfr64
Olfr65
Olfr631
808:ER4 ATAC
809:ER4 ATAC
orig protocol0 _
0.1 _
810: ATAC-seq on G1E-ER4+E2 cells Tn3 50,000 cells PCRn
0_
0.1 _
811: ATAC-seq on G1E-ER4+E2 cells Tn2 100,000 cells PCRn
0_
0.1 _
812: ATAC-seq on G1E-ER4+E2 cells Tn2 150,000 cells PCRn
0_
0.1 _
813: ATAC-seq on G1E-ER4+E2 cells Tn2 50,000 cells PCR3n
0_
0.1 _
814: ATAC-seq on G1E-ER4+E2 cells Tn2 50,000 cells PCR6n
0_
0.1 _
656: ATAC-seq on G1E-ER4+E2 cells
0_
0.1 _
657: ATAC-seq on G1E-ER4+E2 cells run 52
810:ER4 ATAC
3 ml Tn
811:ER4 ATAC
812:ER4 ATAC
813:ER4 ATAC
PCR +(n+3)
814:ER4 ATAC
656:ER4 ATAC
657 run52:ER4 ATAC
7/12/20160 _
8
Repeated ATAC-seq on all 8 progenitor cells
• Increase Tn to 3 ml
• Perform 3 additional cycles of PCR amplification
7/12/2016
+(n+3)
9
Gata2 activated early
Window Position
chr6: 88,060,000
Rpn1
AK194371
knCRMsEry
0.4 841+:HSC ATAC
88,070,000
88,080,000
Mouse July 2007 (NCBI37/mm9) chr6:88,051,001-88,181,000 (130,000 bp)
88,090,000 88,100,000 88,110,000 88,120,000 88,130,000 88,140,000 88,150,000
Gata2
88,160,000
88,170,000 88,180,000
Dnajb8
1700031F10Rik
0_
0.4 849+:HSC ATAC
0_
0.4 842+:CMP ATAC
0_
0.4 850+:CMP ATAC
0_
0.4 843+:GMP ATAC
0_
0.4 851+:GMP ATAC
0_
0.4 844:MEP ATAC
0_
0.4 852:MEP ATAC
0_
0.4 847:CFU-M ATAC
0_
0.4 855:CFU-M ATAC
0_
0.4 848:Megs ATAC
0_
0.4 856:Megs ATAC
0_
0.4 845:CFU-E ATAC
0_
0.4 853:CFU-E ATAC
0_
0.4 846:Ery ATAC
0_
0.4 854:Ery ATAC
7/12/2016 0 _
10
Gata1 activated in CMP, increase in CFU-E
Window Position
chrX:
Hdac6
knCRMsEry
0.8 841+:HSC ATAC
7,530,000
7,535,000
7,540,000
Gata1
Mouse July 2007 (NCBI37/mm9) chrX:7,523,001-7,583,000 (60,000 bp)
7,545,000
7,550,000
7,555,000
7,560,000
7,565,000
7,570,000
7,575,000
7,580,000
Glod5
0_
0.8 849+:HSC ATAC
0_
0.8 842+:CMP ATAC
0_
0.8 850+:CMP ATAC
0_
0.8 843+:GMP ATAC
0_
0.8 851+:GMP ATAC
0_
0.8 844:MEP ATAC
0_
0.8 852:MEP ATAC
0_
0.8 847:CFU-M ATAC
0_
0.8 855:CFU-M ATAC
0_
0.8 848:Megs ATAC
0_
0.8 856:Megs ATAC
0_
0.8 845:CFU-E ATAC
0_
0.8 853:CFU-E ATAC
0_
0.8 846:Ery ATAC
0_
0.8 854:Ery ATAC
7/12/2016 0 _
11
Pf4 and flanking genes activated early, stay on in MEG
Window Position
chr5:
91,150,000
Mouse July 2007 (NCBI37/mm9) chr5:91,110,001-91,265,000 (155,000 bp)
91,200,000
Cxcl5
Pf4
Cxcl3
Ppbp
Cxcl15
91,250,000
knCRMsEry
0.4 841+:HSC ATAC
0_
0.4 849+:HSC ATAC
0_
0.4 842+:CMP ATAC
0_
0.4 850+:CMP ATAC
0_
0.4 843+:GMP ATAC
0_
0.4 851+:GMP ATAC
0_
0.4 844:MEP ATAC
0_
0.4 852:MEP ATAC
0_
0.4 847:CFU-M ATAC
0_
0.4 855:CFU-M ATAC
0_
0.4 848:Megs ATAC
0_
0.4 856:Megs ATAC
0_
0.4 845:CFU-E ATAC
0_
0.4 853:CFU-E ATAC
0_
0.4 846:Ery ATAC
0_
0.4 854:Ery ATAC
7/12/2016
0_
12
Hbb locus shows major activation in CFU-E
Window Position
chr7:
110,950,000
Olfr67
Hbb-b2
Mouse July 2007 (NCBI37/mm9) chr7:110,928,001-111,068,000 (140,000 bp)
111,000,000
Hbb-b1
Hbb-bh2ps
Hbb-y
Olfr66
Hmg14ps
Hbb-bh1
Hbb-bh3ps
Hbb-hb0part
111,050,000
Olfr64a
Olfr64b
Olfr65
Olfr631
knCRMs
3841+:HSC ATAC
0_
3849+:HSC ATAC
0_
3842+:CMP ATAC
0_
3850+:CMP ATAC
0_
3843+:GMP ATAC
0_
3851+:GMP ATAC
0_
3844:MEP ATAC
0_
3852:MEP ATAC
0_
3847:CFU-M ATAC
0_
3855:CFU-M ATAC
0_
3848:Megs ATAC
0_
3856:Megs ATAC
0_
3845:CFU-E ATAC
0_
3853:CFU-E ATAC
0_
3846:Ery ATAC
0_
3854:Ery ATAC
7/12/2016
0_
13
Quantifying peak strength across ATAC-seq datasets
Take the top 100K peaks from each replicate and cell type (2*8=16 datasets).
Peaks tended to be ~500bp. The medians of the distributions for the 16 datasets
ranged from 325-611 bp
Union of all 16 sets of 100K peaks, then merge, keeping the interval that includes
all overlapping peaks.
This step yielded about 360K merged peaks (median = 412 bp)
Compute average mapped read count for each merged peak interval in each
dataset (bigWigAverageOverBed)
Quantile normalize read counts
PCA on scaled read counts
Remove identical read counts
7/12/2016
Compute Pearson’s correlation
coefficient among all pairs of
datasets and plot heatmap
14
Correlations
and hierarchical
clustering based
on ATAC-seq
7/12/2016
15
PCA based on ATAC-seq
12% of variation
7/12/2016
32% of variation
16