Kun:LabNotes/SingleCellExpr/2011-5-11
Jump to navigation
Jump to search
DNA accessibility assay on small numbers of cells[edit]
Idea[edit]
- Performed Nextera tagmentation on cell lysates with and without proteinase digestion. The idea is that for cells that lysed with mild buffer, histones are still attached to the DNA and most likely prevent tagmentation. Therefore, the presence of tagmentation sites on the genome would indicate regions free of histones.
Experiment protocol[edit]
- Make 10ml 10X lysis buffer (LB, 100mM Tris.Hcl pH 7.5, 100mM NaCl, 30mM MgCl2, 1% NP40, Crawford et al. PNAS 2003).
- Prepare aliquots of lymphocytes (GM20431) that contain 1000 cells, 500 cells and 100 cells in 1ul.
- Prepare 2X LB from 10X buffer with and without Qiagen protease (0.1AU/ml). Add one drop of mineral oil.
- Set up another group of controls using another lysis buffer(LB2) that I used on 8/18/2010 (20M EDTA, 10mM Tris.HCl , 200mM NaCl, 0.2% Triton X-100, 0.1AU/ml Qiagen Protease), which could remove DNA binding proteins more completely.
- Mix 1ul cell solution with 1ul lysis buffer
A B C D E F G H I J K 500 cells 200 cells 100 cells 500 cells 200 cells 100 cells NTC 500 cells 200 cells 100 cells NTC cells 1ul 1ul 1ul 1ul 1ul 1ul 1ul(PBS) 1ul 1ul 1ul 1ul(PBS) 2x LB 1ul 1ul 1ul - - - - - - - - 2x LB (Protease) - - - 1ul 1ul 1ul 1ul - - - - LB2 - - - - - - - 1ul 1ul 1ul 1ul
- Incubate at 37C for 30min. For D-K, proceed with heat inactivation at 75C for 15min.
- Perform tagmentation.
Dilute the enzyme: 1:10 For each rxn, used mix of: 1ul 5x LMW Buffer 2ul cell lysate 1ul diluted enzyme 2ul H2O ----------------------- 5ul total / reaction 55C 10 min
- Set up PCR
Tagmentation reaction 10ul KAPA QPCR mix 25ul Orange Primer (10uM) 1ul Blue Primer (10uM) 1ul Bst Pol (5U/ul) 0.5ul H2O 13ul
Orange primer CCTTGCCAGCCCGCTCAG 18nt Blue primer CCTCCCTCGCGCCATCAG 18nt
65C 3min -> 95C 30 sec -> (95C 10sec -> 58C 30 sec -> 72 1min) x 25 -> 72C 3min. Monitor the reactions on a real-time thermal cycler and terminate them before the curves reach saturation.
- Results: See Alice's wiki page.
File:Tagmentation PCR size distribution.png
- The new lysis buffer with NP40 works significantly better than the old buffer with Triton X-100 in making gDNA available for tagmentation. This is a good lysis buffer for genome/exome sequencing on small numbers of cells.
- Protease does make a big difference in terms of DNA accessibility. There is a good possibility that this nuclease accessibility assay would work.
- Real-time PCR signals are dominated by primer-primer interactions, so we can't use the curves to judge the relative amplification efficiencies or determine the amount of input DNA.
- Can we design a Alu-based Taqman or Molecular Beacon primer?
- Alice will purify the amplicons with AmPure beads, then perform the 2nd round PCR to add barcodes and Illumina adaptors.
- In addition to sequencing 9-plex libraries (three protocols x three level of input cells), Alice will also perform exome capture and sequencing on pooled E/F/G libraries. This will tell us whether exome sequencing on 100 cells is feasible.
Sequencing[edit]
- Alice helped me to add barcoded adaptors to the amplicons, and pooled six amplicons with six other methylation samples.
- One lane of 36bp HiSeq reads was generated.
Index Sample Indx1 gentle-lysis-500-cell Indx2 gentle-lysis-200-cell Indx3 gentle-lysis-100-cell Indx4 lyse+protease-500-cell Indx5 lyse+protease-200-cell Indx6 lyse+protease-100-cell
Data analysis[edit]
- Mapping: fastq2bamHg19.pl.
After the mapping, clonal reads were removed such that all the following analyses are based on unique insertion event. /home/kunzhang/softwares/samtools-latest/samtools rmdup -s lane1_Indx2.fastq.bowtie.sorted.bam lane1_Indx2.fastq.bowtie.unique.bam & /home/kunzhang/softwares/samtools-latest/samtools rmdup -s lane1_Indx3.fastq.bowtie.sorted.bam lane1_Indx3.fastq.bowtie.unique.bam & /home/kunzhang/softwares/samtools-latest/samtools rmdup -s lane1_Indx4.fastq.bowtie.sorted.bam lane1_Indx4.fastq.bowtie.unique.bam & /home/kunzhang/softwares/samtools-latest/samtools rmdup -s lane1_Indx1.fastq.bowtie.sorted.bam lane1_Indx1.fastq.bowtie.unique.bam & /home/kunzhang/softwares/samtools-latest/samtools rmdup -s lane1_Indx5.fastq.bowtie.sorted.bam lane1_Indx5.fastq.bowtie.unique.bam & /home/kunzhang/softwares/samtools-latest/samtools rmdup -s lane1_Indx5.fastq.bowtie.sorted.bam lane1_Indx5.fastq.bowtie.unique.bam & /home/kunzhang/softwares/samtools-latest/samtools rmdup -s lane1_Indx6.fastq.bowtie.sorted.bam lane1_Indx6.fastq.bowtie.unique.bam &
Index | Total reads | # Uniquely mapped | # Failed to map | # Non-uniquely mapped | % Uniquely mapped | % Failed to map | % Non-uniquely mapped | Unique non-clonal reads |
Indx1 | 13,118,594 | 8,497,893 | 1,158,048 | 3,462,653 | 64.8% | 8.8% | 26.4% | 879,165 |
Indx2 | 4,749,299 | 3,043,043 | 360,617 | 1,345,639 | 64.1% | 7.6% | 28.3% | 308,255 |
Indx3 | 4,056,368 | 1,748,590 | 1,290,036 | 1,017,742 | 43.1% | 31.8% | 25.1% | 193,567 |
Indx4 | 5,276,661 | 3,599,275 | 309,376 | 1,368,010 | 68.2% | 5.9% | 25.9% | 3,405,291 |
Indx5 | 4,537,535 | 2,988,612 | 349,478 | 1,199,445 | 65.9% | 7.7% | 26.4% | 2,830,296 |
Indx6 | 3,348,382 | 2,076,694 | 268,505 | 1,003,183 | 62.0% | 8.0% | 30.0% | 1,922,737 |
- Comparing insertion sites with ENCODE DNase I hypersensitive data: To test the hypothesis that Tn5 transposon preferentially inserted into genomic regions with high DNA accessibility, I identified all unique insertion sites, and asked whether these sites tend to have higher DNase I accessibility based on the reference data generated by UW and Duke on the lymphocyte cell line GM12878. I downloaded the following three reference data sets from UCSC:
wgEncodeUwDnaseGm12878PkRep1.narrowPeak.txt UW wgEncodeUwDnaseGm12878HotspotsRep1.broadPeak.txt UW wgEncodeOpenChromDnaseGm12878Pk.narrowPeak.txt Duke
For each unique insertion site, I have a accessibility score, which is the 8th column in the BED files. If a site is not mapped to a peak, the score is assigned as zero. Then I load the data into R, and plot the histogram of the scores. I can use Indx4-6 as the controls since all histones and other DNA binding proteins were removed by protease prior to tagmentation.
R commands to load the data and plot the distribution: indx1.np2=read.table("lane1_Indx1.insSite.Gm12878Pk.narrowPeak.txt"); indx2.np2=read.table("lane1_Indx2.insSite.Gm12878Pk.narrowPeak.txt"); indx3.np2=read.table("lane1_Indx3.insSite.Gm12878Pk.narrowPeak.txt"); indx4.np2=read.table("lane1_Indx4.insSite.Gm12878Pk.narrowPeak.txt"); indx5.np2=read.table("lane1_Indx5.insSite.Gm12878Pk.narrowPeak.txt"); indx6.np2=read.table("lane1_Indx6.insSite.Gm12878Pk.narrowPeak.txt"); indx1.np2.hist=hist(indx1.np2$V3,plot=FALSE) indx2.np2.hist=hist(indx2.np2$V3,plot=FALSE) indx3.np2.hist=hist(indx3.np2$V3,plot=FALSE) indx4.np2.hist=hist(indx4.np2$V3,plot=FALSE) indx5.np2.hist=hist(indx5.np2$V3,plot=FALSE) indx6.np2.hist=hist(indx6.np2$V3,plot=FALSE) plot(indx4.np2.hist$mids, log10(indx4.np2.hist$density),type="p",col="blue",xlab="score",ylab="log10(density)") points(indx5.np2.hist$mids, log10(indx5.np2.hist$density),type="p",col="green") points(indx6.np2.hist$mids, log10(indx6.np2.hist$density),type="p",col="red") lines(indx1.np2.hist$mids, log10(indx1.np2.hist$density),col="blue") lines(indx2.np2.hist$mids, log10(indx2.np2.hist$density),col="green") lines(indx3.np2.hist$mids, log10(indx3.np2.hist$density),col="red")
Overlap with UW GM12878 Hotspots BroadPeak Overlap with UW GM12878 NarrowPeak Overlap with Duke GM12878 BroadPeak File:Overlap with GM12878 HotspotsRep1.broadPeak.png File:Overlap with GM12878 PkRep1.narrowPeak.png File:Overlap with GM12878 Pk.narrowPeak.png Lines: gentle lysis; Circles: lysis + protease digestion
- Conclusion:
- It appeared that there was an enrichment in Indx1-3 for highly accessible regions in the genome.
- Discussion.
- I initially did the mapping to hg18 because I thought the reference data were based on hg18, and did not see any enrichment. That's another indication that the enrichment is real.
- There were high percentages of clonal reads. In the future experiments, it is a good idea to increase the reaction time for the tagmentation reactions, such that we can saturate the accessible regions with transposon integration.
- Regions with one single insertion sites won't be amplified by PCR. However, such insertion events are informative. Can we figure out a way to capture such insertion sites? One idea is that we can use transposon to add T7 promoter sequence, such that we can do amplification by in vitro transcription.
- Another informative exercise would be to download the raw reads for one of the reference data set from UW or Duke, down-sample to a similar level as Indx1-3, and repeat the same enrichment analysis. This will give us an idea on how close our assay is to existing methods.
- In addition, we can also compare with the reference data generated on other lymphocytes lines to get an idea about the variability among different lymphocyte lines.