Dinh 2011/NOTES/2011-10-7

From ZhangLabWiki
Jump to navigation Jump to search

To-Do for Nature Methods revision on "Library-free Methylation Sequencing with Bisulfite Padlock Probes"[edit]

Draft by Thurs, Oct 20, 2011

From Revision Document:

  1. LC Sciences Oligo Experiment - Noi
        1. Order primers for probe preparation - Dinh File:LC Sciences DMR Probes Design.pdf
  3. Package BisReadMapper correctly - Dinh
        1. Include SNP filtering script
        2. Make sure there are no errors in code
        ChangeNote:
        September 12, 2011
        #genomePrep.pl to stream-line reference genome preparation
        #bisReadMapper to read in chromosome sizes
        October 13, 2011
        #set all parameters with params.txt file File:SampleParams.txt
        #samtools mpileup output works with pileup2variantsStat
        #generate a separate C-SNP bed file
        #option to call CHH and CHG methylation         
        3. Make sure installation package works on a fresh computer
        4. Include a better manual on how to use the program File:BisReadMapper v1.2manual.pdf
  4. Distribution of QC Metric within and between batches of samples- Dinh
        Within samples, plot Watson vs Crick for H1, PGP1iPS, PGP1F in one figure
        Within a batch - PGP1iPS-1 vs PGP1iPS-2
        Between batch - ask Noi if current run has any
        1. Mention in main text
        2. Create supplementary figure or table
        Within samples:
        PGP1iPS
        N=44361 r=0.959125450182116
        PGP1F
        N=55965 r=0.96136269717601
        H1
        N=29884 r=0.966980865585974
       Within one batch:
       PGP1iPS_1vs2
       N=405508        r=0.979670718560575
       Between processed batches *new data only got 1/2 of target, so N should be about double the value shown*
       PGP1iPS_betweenBatches
       N=117186        r=0.965899105841211


  5. Analysis of capture performance - Dinh
        Use libraries prepared by MmeI protocol
        1. # Unique CpGs targeted (overall) = 10,658,852
        2. Sensitivity, Specificity, and Uniformity of Capture - 
             Sensitivity - File:Cpg330kEfficiency1sample.png
             Specific - % on/off target CpGs
             Uniformity - Show VennDiagram between batches number of CpG's unique and shared between batches
        3. Off-target sequencing/capture vs on-target
             Calculate reads mapped to target regions and those that don't
             Using trimmed.cAWS1_330k.fastq.*.sorted.bam files
             Reverse strand:
             Reads = 13848999         On Target = 11702798 (0.845)
             Reads mapping to multiple targets = 637943  (0.046)
             Forward strand:
             Reads = 16112051         On Target = 12951607 (0.804)
             Reads mapping to multiple targets = 921670  (0.057)
             Summary: 0.874936559 of reads mapped are on target.
        4. Capture efficiency for each probe, subsetting/suppressors
             This is tied to capture sensitivity above
        5. Venn diagram illustrating overlap in coverage between two samples (H1 and PGP1F, for example)
  6. SNP Calilng in bisReadMapper - Dinh
        Overlap PGP1F, PGP1L, and PGP1iPS data for 2-3 overlaps to get a set of reliable SNPs, calculate % match of PGP1F calls with bisulfite data.
        1. Mention that we can do this in main text briefly, and mention advantages (built-in genotyping)
        2. Add section to supplementary information on how this is performed; include some accuracy statistics on PGP1 and CV
        File:PGP1F OverlapbisSNPvsCGI variants.txt - 13227/13999 SNPs matched, 94.5%
        Distribution of miscalled SNPs:
        A/A 3
        A/C 94
        A/G 173
        A/T 48
        C/C 25
        C/G 129
        C/T 169
        G/G 3
        G/T 73
        T/T 55
  7. Clarifications - Everyone
        1. Improve line 79-82 (cis-interactions, long-range interactions)
              1. More clarification about GREAT vs true cis-interactions
        2. Cost analysis
              1. Clarify cost analysis (which costs we are including)
              2. Cost of probe libraries from LC Sciences + cost of sequencing from Biogem
              3. Compare costs to RRBS and WGBS for different numbers of samples and targets ( Representative sample sizes: 10 100, 1000 samples) and for about the same number of target regions we have ~100K (or that of Illumina CpG Array)
              For RRBS - http://www.nature.com/nmeth/journal/v7/n2/full/nmeth.1414.html , http://www.nature.com/nmeth/journal/v7/n2/fig_tab/nmeth.1414_T1.html
              Some numbers on WGBS - http://www.nature.com/nprot/journal/v6/n4/fig_tab/nprot.2010.190_T1.html
        3. Clarify difference between N2 protocol and library-free
              1. Mention advantages and disadvantages of each
              2. Clarify that CpGs covered - make it per megabase sequencing effort
        4. Clarify 100 bp PE vs 110 bp PE - mention CpGs covered per megabase
        5. Clarify statistical significance of DMRs
              1. Clarify that we only tested sites with more than 0.1 abs methylation difference, and that these were not statistically significantly different
              2. Mention that some sites had high absolute difference, but this was not statistically significant
        6. Emphasize the flexibility of our assay - we can target many different regions of interest
        7. Emphasize in bisReadMapper paragraph that we can map data of any bisulfite status in an unbaised manner - redo diagram to show GUESS-STRAND module; emphasize other novel aspects as well (including SNP calling)
  8. Upload correct probe set (H1, H2 sequences only for DMR330K) and target region file (206,312 regions) to website (include LC Sci probes and target region set) - Dinh