Dinh/Dinh 2015/NOTES/2015-8-27

From ZhangLabWiki
Jump to navigation Jump to search

BSPP Capture Experiment - Tumor DNA & Buffy Coat from Cambridge[edit]

  • 1/2 HiSeqRapidRun data are copied to: /media/Ext12T/DD_Ext12T/MONOD/150828_HiSeqRapidRun
  • 72 samples from Cambridge
  • 8 GM12878 + Panc1 gDNA mixtures: 1%,5%,10%,20%,40%,60%,80%,100% (% Panc1 added)

Bisulfite conversion with Zymo EZ-MagPrep 96 wells kit[edit]

  • Loss a majority of the DNA with the bisulfite conversion. I believe this is due to overdrying. Noi's protocol deviated from the Zymo protocol in the drying time, she only used 10 minutes to dry. I also think she might have been better at removing ethanol than Dana and I. Next time, we should use better techniques and p10 to remove as much ethanol as possible.

Quantification of converted DNA[edit]

qPCR with ActB promoter primers[edit]

  • First time with Kapa Hifi Uracil got step-ladder but mostly smear on gel. Second time with Kapa Hifi Uracil+ we doubled the input template but got no specific product on the gel. We also didn’t get good correlation between the two replicates so decided that the results are not trustworthy.

QUBIT[edit]

Qubit on row A: mostly out of range (too low) and some between 0.33 ng/uL - 1 ng/uL Qubit on B,D,E,F (cols 2,4,6,8): 50% out of range (too low) and others between 0.210 - 1.2 ng/uL. Maximum yield was around 22 ng and minimum yield could be < 1 ng. Majority of samples are around 4-10 ng.

June2015 MONOD - Capture[edit]

  • We carried on with capture, assuming 20 ng input across the board. We used 70 ng input for the mixtures which got more yield. We prepared capture with 500:1 probe to target ratio, 100:1 suppressor to probe ratio, and hybridization temperature of 55C.
  • Program:
1. 95°C    30s
2. 94.8°C  30s
-0.2°C per 30s 198 times
3. 55°C    20h 
ADD 4.3 uL KLN mix
4. 55°C    20h
5. 94°C    2m
ADD 1.5 uL ExoI/ExoIII (10 U ExoI and 100 U Exo III) mix
6. 37°C    2h
7. 90°C    5m
8. 4°C     forever
  • KLN:
AmpLigase: 1U/20uL = 1.5
HemoKlenTaq Polymerase: 5U/20uL = 7.5
To add 4.3 uL of KLN, final concentrations must be:
Polymerase: 1.75 U/uL
AmpLigase: 0.35 U/uL
dNTP: 0.6 uM for 10 ng or 4 uM for 70 ng
AmpLigase Buffer 1X 
4 uL of KLN mixture is:
2.8 uL of 2.5U/uL HemoKlentaq (NEB)
0.28 uL of 5U/uL AmpLigase (Epicentre)
0.52 uL of 4.6 uM or 32.2 uM dNTPs (NEB)
0.40 uL of 10X Ampligase Buffer (Epicentre)

Capture test PCR (use 1.6 uL of capture reactions)[edit]

  • Reaction set up:
Phusion HS Flex (Fishers Scientific)     5 uL
Circularized DNA                                  1.6 uL
AmpFSol6.4 (10 uM)                             0.2 uL
AmpRIndXXSol (10 uM)                        0.2 uL
Nuclease free water                               3.0 uL
  • Program:
98C (10s) -> [ 98C (10s) -> 58C (30s) -> 72C (20s) ] - 8 cycles -> [ 98C (10s) -> 72C (20s) ] - 12 cycles -> 72C (3 min)
  • From the test capture gel images, we picked 20 tumor/buffy coat samples to continue with library preparation. Other samples were on the border of being good, so we decided to hold off on them until later.

Capture PCR (use 30 uL of capture reactions)[edit]

  • Reaction set up:
Phusion HS Flex (Fishers Scientific)     50 uL
Circularized DNA                                  30 uL
AmpF_V7 (10 uM)                                    2 uL
AmpR_V7 (10 uM)                                    2 uL
Nuclease free water                               16 uL
  • Program:
98C (10s) -> [ 98C (10s) -> 58C (30s) -> 72C (20s) ] - 8 cycles -> [ 98C (10s) -> 72C (20s) ] - 12 cycles -> 72C (3 min)
  • Scaled up the amount of capture reactions per 100 ul PCR reaction to 30 uL. Phusion HotStart Flex master mix seems to be able to handle this.
  • Used the Dual-Indexing primers to prepare the sequencing libraries.

HiSeqRapidRun 8/28/15[edit]

  • We ran these libraries with Chris' capture libraries. While the Read 1 primers were different, they would not cause problems with our libraries. However, Chris's (AmpR6IndSeq) primer for indexing would cause earlier sequencing of my libraries' i7 barcode sequences (about 9 bases early). To resolve this issue, I decided to only use the AmpR6IndSeq primer, and ask for 17 total indexing cycles so that it would finish reading my i7 barcodes.
  • We got 360 million SE reads from this run. The Dual Index libraries got 120,656,345 reads.

Data analysis[edit]

Mapping statistics and QC[edit]

  • don't have enough reads at 50X for f/r correlation
LabWellID Sample description i7 i7_index i5 i5_index Total SE reads Total reads after adaptor and quality trimming % removed after trimming Total mapped %mapped Total UMIs %clonal Conversion rate
1B tumour N726 CCTAAGAC S513 TCGACTAG 2,726,281 2,674,529 1.9% 580,434 21.7% 94,468 84% 98.1%
1C tumour N726 CCTAAGAC S515 TTCTAGCT 3,085,341 3,042,098 1.4% 697,995 22.9% 137,110 80% 98.0%
1F tumour N726 CCTAAGAC S516 CCTAGAGT 10,560,939 4,889,796 53.7% 873,152 17.9% 78,107 91% 97.7%
2A tumour N726 CCTAAGAC S517 GCGTAAGA 3,950,484 3,901,822 1.2% 928,741 23.8% 166,564 82% 98.4%
2C tumour N726 CCTAAGAC S518 CTATTAAG 3,793,447 3,762,701 0.8% 1,008,596 26.8% 208,873 79% 98.6%
2D tumour N726 CCTAAGAC S520 AAGGCTAT 3,891,246 3,840,554 1.3% 980,524 25.5% 203,284 79% 98.8%
2E tumour N726 CCTAAGAC S521 GAGCCTTA 5,247,942 5,004,681 4.6% 945,710 18.9% 83,356 91% 97.6%
2F tumour N726 CCTAAGAC S522 TTATGCGA 2,888,588 2,854,629 1.2% 612,504 21.5% 104,696 83% 97.8%
2G tumour N727 CGATCAGT S513 TCGACTAG 4,315,801 4,290,600 0.6% 1,372,394 32.0% 379,643 72% 96.9%
2H tumour N727 CGATCAGT S515 TTCTAGCT 3,735,343 2,881,322 22.9% 737,836 25.6% 107,416 85% 97.7%
3E tumour N727 CGATCAGT S516 CCTAGAGT 2,377,407 2,304,657 3.1% 443,741 19.3% 58,580 87% 98.9%
4A tumour N727 CGATCAGT S517 GCGTAAGA 3,253,122 2,723,420 16.3% 486,178 17.9% 56,921 88% 98.6%
4D tumour N727 CGATCAGT S518 CTATTAAG 3,483,350 3,281,562 5.8% 585,217 17.8% 84,904 85% 99.1%
4H tumour N727 CGATCAGT S520 AAGGCTAT 3,215,302 2,801,458 12.9% 595,286 21.2% 73,622 88% 98.8%
5E tumour N727 CGATCAGT S521 GAGCCTTA 4,520,578 4,434,329 1.9% 1,037,273 23.4% 182,801 82% 98.6%
5G tumour N727 CGATCAGT S522 TTATGCGA 4,200,131 4,142,399 1.4% 1,145,682 27.7% 229,900 80% 98.1%
8B tumour N728 TGCAGCTA S513 TCGACTAG 11,879,016 3,471,904 70.8% 392,825 11.3% 44,429 89% 98.8%
8H buffy coat N728 TGCAGCTA S515 TTCTAGCT 5,094,273 4,441,419 12.8% 641,792 14.5% 94,633 85% 98.8%
9A buffy coat N728 TGCAGCTA S516 CCTAGAGT 2,757,712 2,644,256 4.1% 531,175 20.1% 191,320 64% 96.3%
9H buffy coat N728 TGCAGCTA S517 GCGTAAGA 3,672,784 3,095,587 15.7% 412,545 13.3% 57,614 86% 98.9%
M1 1% Panc1 N729 TCGACGTC S513 TCGACTAG 5,524,494 4,537,156 17.9% 765,599 16.9% 194,514 75% 98.6%
M2 5% Panc1 N729 TCGACGTC S515 TTCTAGCT 3,622,228 3,458,155 4.5% 546,327 15.8% 202,248 63% 98.5%
M3 10% Panc1 N729 TCGACGTC S516 CCTAGAGT 3,974,239 3,796,675 4.5% 946,653 24.9% 216,026 77% 98.5%
M4 20% Panc1 N729 TCGACGTC S517 GCGTAAGA 3,996,449 3,607,383 9.7% 553,868 15.4% 218,986 60% 98.2%
M5 40% Panc1 N729 TCGACGTC S518 CTATTAAG 2,308,086 2,262,764 2.0% 565,189 25.0% 210,432 63% 98.6%
M6 60% Panc1 N729 TCGACGTC S520 AAGGCTAT 7,451,464 7,012,189 5.9% 649,500 9.3% 336,409 48% 98.5%
M7 80% Panc1 N729 TCGACGTC S521 GAGCCTTA 3,724,922 3,552,045 4.6% 611,140 17.2% 249,179 59% 98.4%
M8 100% Panc1 N729 TCGACGTC S522 TTATGCGA 1,405,376 1,062,567 24.4% 150,040 14.1% 54,736 64% 98.3%

Quantification compared with RRBS[edit]

  • Panc1 RRBS data vs M8 (100% Panc1 data): Pearson's R = 0.907 at 20X

File:Dinh BSPP vs RRBS panc1SmoothScatter.png

Titration series with Mixtures[edit]

  • Does the methylation levels correlate well with titration amounts?
  • Median Pearsons correlation of methylation value with linear regression model value shown on graph
  • Choose only sites with minimum 20X coverage.

File:Dinh MONOD June2015 titrationSeries Panc1.png

Samples hierarchical clustering[edit]

  • Nice to see that the 3 buffy coats (8H, 9A, 9H are clustered together).
  • Also large methylation differences. This graph was generated allowing 5/20 missing values. May need to repeat with no missing value.

File:Dinh MONOD June2015 tumourOvarianCancer dana.png

BRCA1 methylation[edit]

  • Do the tumor samples and buffy coat samples cluster separately?
  • Genome browser

File:Dinh MONOD June2015 ovarian cancer brca1 genomebrowser.png