Ns126:Calendar/NOTES/2016-2-1

From ZhangLabWiki
Jump to navigation Jump to search

Cancer Specific Methylation haplotype region show increased MHL in the plasma

  • From Dr. Zhang
  • I think I can rationalize the difficulty in detecting tumor haplotypes in plasma at a sufficient high level of sensitivity and specificity.
  • We are sequencing the plasma samples with very limited input DNA, equivalent to only a few hundred copies per sample. With the additional major loss of material during bisulfite conversion and library construction, we are probably sampling one or two dozens of molecules per sample. On the other hand, the fraction of tumor DNA in plasma is typically in the range of 0.1% or lower. So if there are 1,000 regions that has tumor-specific haplotypes at 0.1% level and we are looking at 10 unique molecules per region, we would expect to see one tumor haplotypes in only 10 of 1,000 regions. If you move to the next sample, you would see another 10 regions. The chance of finding a set of common regions where we can consistently detect tumor haplotypes is extremely low, at least at the sample size of <100 that we have.
  • Along this line of thoughts, random forest is not going to be the right approach for such data. Our best bet is to compare data from tumor tissues and normal plasma, identify a set of genomic regions that are clean in plasma and contain certain level of methylated haplotypes in tumor, then ask whether these regions have an elevated level of MHL.
  1. Dr. Zhang's Round 1 analysis: http://genome-tech.ucsd.edu/LabNotes/index.php/Kun:LabNotes/MONOD/2014-8-12#5._Searching_for_primary_tumor_specific_haplotypes
  2. Merge BAM files of primary cancer, cancer plasma and normal plasma
  3. LMS screening in normal plasma samples.
  4. Identify tumor-specific haplotype regions (primary solid tissue vs normal plasma)
  5. Calculate the cumulation/average MHL for these region (tumor specific) for a plasma sample.
  6. Then assign a probability for the plasma is belong to 'cancer' or 'normal' as well as which tissue-of-origin?
  7. A panel of canonical cancer-related methylation region, and a random genomic region as control)?

Round 1: MHL in imputed,normalized merge data which prediction analysis missed

  • 1, Find out regions that contain HMHs in primary tumor but not in N37 normal tissues or normal plasma.
scp shg047@genome-miner.ucsd.edu:/home/kunzhang/CpgMIP/MONOD/Data/
cat PC-P-*.methHapCounts.txt | ../scripts/get_tumor_specific_HMH_regions.pl N37_ten_tissues_NC-P-ALL_RRBS-dRRBS.methHapCounts.txt  > PC-P_N37_plasma_specific_HMH_regions.bed &
cat 7-P-*.methHapCounts.txt | ../scripts/get_tumor_specific_HMH_regions.pl N37_ten_tissues_NC-P-ALL_RRBS-dRRBS.methHapCounts.txt  > LC-P_N37_plasma_specific_HMH_regions.bed &
cat 6-P-*.methHapCounts.txt | ../scripts/get_tumor_specific_HMH_regions.pl N37_ten_tissues_NC-P-ALL_RRBS-dRRBS.methHapCounts.txt  > CRC-P_N37_plasma_specific_HMH_regions.bed &
samtools view -H 6-P-10.sorted.clipped.bam > header.sam
samtools merge -h header.sam 6-P-10.sorted.clipped.bam 6-P-1.sorted.clipped.bam 6-P-2.sorted.clipped.bam 6-P-3.sorted.clipped.bam 6-P-4.sorted.clipped.bam 6-P-5.sorted.clipped.bam 6-P-6.sorted.clipped.bam 6-P-7.sorted.clipped.bam 6-P-8.sorted.clipped.bam 6-P-9.sorted.clipped.bam  
  • 1,define clean genomic region in normal plasma (low methylation region)

704 haplotype regions were identified as the cancer specific(solid tissue) and were clean in the normal plasma samples. File:Figure 4D.txt

  • 2 Calculate the cumulation/average MHL for these region (tumor specific) for a plasma sample.
    • MHL distribution for above defined regions

File:7AB3.tm.png

      • CP: cancer plasma, CT: cancer tissue, NP: normal plasma
  • 3. assign a probability for the plasma is belong to 'cancer' or 'normal' as well as which tissue-of-origin?
Colon: 740	
> myData	
  type        mean         sd   n          sem           me	
1   CP 0.303570259 0.04889567 740 0.0184808266 2.149566e+00	
2   NP 0.001489827 0.00036459 740 0.0001378021 9.520868e+11	
3   TP 0.004369450 0.01452305 740 0.0054891956 2.759810e+00	
 
Lung: 740	
> myData	
  type        mean         sd   n          sem           me	
1   CP 0.310171625 0.05194363 740 0.0196328452 2.137493e+00	
2   NP 0.001489827 0.00036459 740 0.0001378021 9.520868e+11	
3   TP 0.008964632 0.02154434 740 0.0081429953 2.444373e+00	
 

Round 2: Tunor specific Haplotype regions called by Dr.Zhang's script

  • Dinh shared her alignment Bam file
  • Calling haploinfo and identify cancer specific hapltype region with Dr.Zhang's previous script
  • Change name to unified format 6-P, 7-P, PC-P code
# merge hapInfo
cd /home/shg047/monod/hapinfo
cat /home/shg047/monod/rrbs_kun/6-P-*.hapInfo.txt > ../mixHap/6-P-ALL.hapInfo.txt
cat /home/shg047/monod/rrbs_kun/7-P-*.hapInfo.txt > ../mixHap/7-P-ALL.hapInfo.txt
cat /home/shg047/monod/rrbs_kun/PC-P-*.hapInfo.txt > ../mixHap/PC-P-ALL.hapInfo.txt
cat /home/shg047/monod/rrbs_kun/NC-P-*.hapInfo.txt > ../mixHap/NC-P-ALL.hapInfo.txt
cat /home/shg047/monod/rrbs_kun/6-T-*.hapInfo.txt > ../mixHap/6-T-ALL.hapInfo.txt
cat /home/shg047/monod/rrbs_kun/7-T-*.hapInfo.txt > ../mixHap/7-T-ALL.hapInfo.txt
cat /home/shg047/monod/rrbs_kun/PC-T-*.hapInfo.txt > ../mixHap/PC-T-ALL.hapInfo.txt
# find NP-P-high LOD target
cd /home/shg047/monod/mixHap
perl /home/shg047/monod/bin/mixMethHapAnalysis_19Aug14.pl  NC-P-ALL.hapInfo.txt 6-T-ALL.hapInfo.txt 6-P-ALL.hapInfo.txt /home/kunzhang/CpgMIP/MONOD/Public_data/WB_middle_age_UMR.BED.txt |sort -k 2,2nr > RRBS_target_LOD_CC.txt 
perl /home/shg047/monod/bin/mixMethHapAnalysis_19Aug14.pl  NC-P-ALL.hapInfo.txt 7-T-ALL.hapInfo.txt 7-P-ALL.hapInfo.txt /home/kunzhang/CpgMIP/MONOD/Public_data/WB_middle_age_UMR.BED.txt  |sort -k 2,2nr  > RRBS_target_LOD_LC.txt 
perl /home/shg047/monod/bin/mixMethHapAnalysis_19Aug14.pl  NC-P-ALL.hapInfo.txt PC-T-ALL.hapInfo.txt PC-P-ALL.hapInfo.txt /home/kunzhang/CpgMIP/MONOD/Public_data/WB_middle_age_UMR.BED.txt |sort -k 2,2nr > RRBS_target_LOD_PC.txt 
./find_NC-P_high-LOD_targets.pl > NC-P_high-LOD_targets.txt
  • Then what about Dr. Zhang's result in Phase 1 Plasma dataset
perl /home/shg047/monod/bin/mixMethHapAnalysis_19Aug14.pl  NC-P-ALL.hapInfo.txt 6-T-ALL.hapInfo.txt 6-P-ALL.hapInfo.txt  /home/shg047/monod/mixHap/zhang/WB_middle_age_UMR.BED.txt| sort -k 2,2nr > /home/shg047/monod/mixHap/zhang/RRBS_targets_LOD_CC.txt 
perl /home/shg047/monod/bin/mixMethHapAnalysis_19Aug14.pl  NC-P-ALL.hapInfo.txt 7-T-ALL.hapInfo.txt 7-P-ALL.hapInfo.txt   /home/shg047/monod/mixHap/zhang/WB_middle_age_UMR.BED.txt| sort -k 2,2nr  > /home/shg047/monod/mixHap/zhang/RRBS_targets_LOD_LC.txt 
perl /home/shg047/monod/bin/mixMethHapAnalysis_19Aug14.pl  NC-P-ALL.hapInfo.txt PC-T-ALL.hapInfo.txt PC-P-ALL.hapInfo.txt   /home/shg047/monod/mixHap/zhang/WB_middle_age_UMR.BED.txt| sort -k 2,2nr > /home/shg047/monod/mixHap/zhang/RRBS_targets_LOD_PC.txt 
cd /home/shg047/monod/mixHap/zhang

Round 3: Tumor specific MHBs in patients plasma: normal tissue (NT), normal plasma (NT), cancer tissue(CT), cancer plasma(CP)

First Time Update: 01/03/2016

  • requirement: mhl<0.1 in normal while mhl>0.5 in cancer tissue. other cut-off were tried, results were similar.
  • lung cancer (3 normal tissue, 5 cancer tissues, 26 normal plasma, 30 lung cancer plasma)
** MHL matrix:  /home/shg047/monod/rrbs_kun/MOND.MHL.txt
** Find tumor specific MHBs and calculate MHL in lung cancer patients plasmas
** lung cancer tumor specific MHBs:File:Lung.mixhap.mhl.in.plsma.txt in /home/shg047/oasis/monod/mixHap/lung.mixhap.mhl.in.plsma.txt
  • colon cancer(3 normal tissue, 5 cancer tissues, 26 normal plasma, 29 colon cancer plasma)
** MHL matrix:  /home/shg047/monod/rrbs_kun/MOND.MHL.txt
** Find tumor specific MHBs and calculate MHL in colon cancer patients plasmas
** colon cancer tumor specific MHBs:File:Colon.mixhap.mhl.in.plsma.txt /home/shg047/oasis/monod/mixHap/colon.mixhap.mhl.in.plsma.txt
  • pancreatic cancer(3 normal tissue, 5 cancer tissues, 26 normal plasma, 10 pancreatic cancer plasma)
** MHL matrix:  /home/shg047/monod/rrbs_kun/MOND.MHL.txt
** Find tumor specific MHBs and calculate MHL in pancreatic cancer patients plasmas
** pranceatic cancer tumor specific MHBs:File:Pancreatic.mixhap.mhl.in.plsma.txt in /home/shg047/oasis/monod/mixHap/pancreatic.mixhap.mhl.in.plsma.txt

Second Time: Update: 10/03/2016

  • Dr. Zhang want to check the heatmap to validate the result is solid. I made the heatmap plot by excel since there are some "NA" in the dataset.
cd /media/LTS_33T/SG_LTS33T/monod/mixHap
Rscript colon.mixhap.plasma.R
Rscript lung.mixhap.plasma.R
Rscript pancreatic.mixhap.plasma.R
File:Matrix PlasmaHighMHL.xlsx
File:Matrix PlasmaHighMHL.xlsx
File:Matrix PlasmaHighMHL.xlsx
  • Achieve the genomic regions to above excel files (bed file)
cd /home/shg047/oasis/monod/mixHap
cut -f 1 colon.data.plsma-hypoall.txt > colon.hypermhl.plasma.txt
cut -f 1 lung.data.plsma.hypoall.txt > lung.hypermhl.plasma.txt
cut -f 1 pancreatic.data.plsma.hypoall.txt > pancreatic.hypermhl.plasma.txt
File:Colon.hypermhl.plasma.txt # 1367
File:Lung.hypermhl.plasma.txt # 784
File:Pancreatic.hypermhl.plasma.txt # 608
  • File:34D5.tm.png
  • Obvously, The MHL for the cancer plasma is not as expected as Dr. Zhang. There are large number of high MHL regions correspond to solid cancer tissues.
  • Then I checked the MHL distribution in all the samples of MONOD project. I found the overall MHL in cancer plasma were higher than some other samples as the following figure shown (click and zoom out)
  • File:887.tm.png
  • Then Dr. Zhang want to collect regions which are hyper in cancer tissue however hypo in all the normal samples (N37+SALK). Again, the MHL was as the following figure:
  • File:45A7.tm.png
  • The boxplot for MHL distribution in 4 categories.
  • File:6697.tm.png

Third Time: Update: 11/03/2016

  • Find 4 loci to check it again.
cd /home/kunzhang/CpgMIP/MONOD/Data/1407-combined_RRBS/mld_blocks_stringent_hapInfo
grep 8054556 7-P-2.mld_blocks_r2-0.5.hapInfo.txt
chr17:8054556-8054584   CCCCC   1       8054556,8054568,8054579,8054581,8054584
cd /home/shg047/monod/hapinfo
grep 8054556 7-P-14.hapInfo.txt
chr17:8054556-8054584   CCCCC   1       8054556,8054568,8054579,8054581,8054584
cd /home/shg047/monod/hapinfo
grep 8054556 7-P-22.hapInfo.txt
chr17:8054556-8054579   CCC     1       8054556,8054568,8054579
cd /home/shg047/monod/hapinfo
grep 95947328 7-P-11.hapInfo.txt
chr9:95947328-95947356  CCCCC   2       95947328,95947337,95947340,95947345,95947356
chr9:95947328-95947356  TTTTT   1       95947328,95947337,95947340,95947345,95947356
cd /home/shg047/monod/hapinfo
grep 100204208 PC-T-2.hapInfo.txt
chr14:100204208-100204257       CCT     1       100204208,100204221,100204232
cd /home/shg047/monod/hapinfo
grep 100204208 PC-P-2.hapInfo.txt
chr14:100204208-100204257       CT      1       100204250,100204257
  • You can try check any postion with the following command:
cd /home/shg047/monod/hapinfo
grep loci hapinfo.file

  • Also You can try to check MHL in hapinfo you created several month ago. These two version MHL don't have big different. I just let longer haplotype as I said before, such as CCCCCCCCC
cd /home/kunzhang/CpgMIP/MONOD/Data/1407-combined_RRBS/mld_blocks_stringent_hapInfo
grep loci hapinfo.file

Round 4: Check the Batch 1 and Batch 2 samples

  • Dr.Zhang found the signficant difference between batch 1 and batch 2 samples.
  • Dinh check the difference between phase I and phase II samples -
    • (1) bisulfite conversion rate (quantified by Lambda DNA spike-in) were between 98.54%-99.40% for phase I and lower between 96.61%-98.63% for phase II.
    • (2) According to Noi's wikinotes, the ChargeSwitch gDNA serum kit was used to extract DNA from plasma in phase I and we used QIAmp plasma nucleic acid kit in phase II.
  • Then how about the methylation level difference between batch 1 and batch 2
    • Select same loci check the methylation level difference between batch 1 and batch 2
    • Use same loci to do the clust analysis.
    • Do we have clinical information to the samples?
    • Dinh found no batch effect between phase 1 and phase 2 as the following.
  • File:32E9.tm.png
  • box-plot for all the samples is as the following
  • File:199.tm.png
  • Density plot is as the following:
  • File:DF.tm.png
  • MHL Difference between Batch 1 and Batch 2 sets
  • File:3573.tm.png