Kun:LabNotes/CpgSeq

From ZhangLabWiki
Jump to navigation Jump to search

2014 <calendar> name=Kun:LabNotes/CpgSeq format=%name/%year-%month-%day date=2014/02/01 view=oneyear </calendar> 2013 <calendar> name=Kun:LabNotes/CpgSeq format=%name/%year-%month-%day date=2013/02/01 view=oneyear </calendar> 2012 <calendar> name=Kun:LabNotes/CpgSeq format=%name/%year-%month-%day date=2012/02/01 view=oneyear </calendar> 2011 <calendar> name=Kun:LabNotes/CpgSeq format=%name/%year-%month-%day date=2011/02/01 view=oneyear </calendar> 2010 <calendar> name=Kun:LabNotes/CpgSeq format=%name/%year-%month-%day date=2010/02/01 view=oneyear </calendar> 2009 <calendar> name=Kun:LabNotes/CpgSeq format=%name/%year-%month-%day date=2009/02/01 view=oneyear </calendar> 2008 <calendar> name=Kun:LabNotes/CpgSeq format=%name/%year-%month-%day date=2008/02/01 view=oneyear </calendar> 2007 <calendar> name=Kun:LabNotes/CpgSeq format=%name/%year-%month-%day date=2007/11/01 view=threemonths </calendar>

Probe design[edit]

Human DMR 460k set (Jan 2014)[edit]

Human DMR 330k set (July 2011)[edit]

Human DMR 220k expansion set (Dec 2010)[edit]

   LibraryName    OLID
   DMR220kExp1   0320561
   DMR220kExp2   0320571
   DMR220kExp3   0320581

Human DMR 220k probe set (May 2010)[edit]

  • This probe set was designed based on the following targets:
    • core:
      • ~20,000 DMRs provided by Andy Feinberg's group at JHU. These DMRs were identified by the CHARM analysis;
      • ~30,000 targets provided John Stamatoyannopoulous's group at UW. These targets were mainly selected based on DNase I sensitive regions.
    • imprinted: Imprinted genes (Robert).
    • h1imr90enh: Enhancers in H1 and IMR90 cell lines (Robert).
    • salkaml: Additional DMRs based on Salk's BisSeq data and AML data (Robert).
  • Parameter file for probe design.
  • Probe assembly script.
  • Probes designed and ordered for each subset:
             Designed  Ordered
    Core.50k  170567   170567
    H1IMr90Enh  8023     8023
    Imprinted   3304     3304
    SalkAml    52080    38106
    Total:    233974   220000
  • All 220k probes were split into four 55k libraries:
  • The probes were re-grouped based on the relative capturing efficiencies:
    • Probe efficiency and grouping info file.
    • DMR220k.norm.s1_6500 0294691
    • DMR220k.norm.s2_6500 0294701
    • DMR220k.norm.s3_55k 0294711
    • DMR220k.norm.s4_55k 0294721
    • DMR220k.norm.s5_55k 0294731
    • DMR220k.norm.s6_27k 0294741
    • DMR220k.norm.s7_27k 0294751

Mouse probes (Apr. 2009)[edit]

   Set ID       # Targets      #probes           
   JHU_mm9          88          8,682
   ChrX_mm9        893         35,291
   ChrY_mm9         15            521
   Imprint_mm9     114          6,786
   miRNA_mm9       493         27,360
   GO_mm9         1170         
  • First probe set (CpgMm9A) ordered:
  • 55,000 probes.
    • JHU_mm9
    • ChrY_mm9
    • Imprint_mm9
    • ChrX_mm9
    • miRNA (first 3720 probes)

Agilent probes, V3 (Feb. 2009)[edit]

  • Targets
    • 16,379 T-DMRs identified by Irizarry et al. Nature Genetics 41:178-86, 2009
    • SwitchGear TSS that overlap with CCDS first exons (-400bp to 200bp relative to TSS). Only 775 targets on chrX were included.
    • 179 known or predicted imprinted genes
  • Degenerate probes: Probe Information probe2padlockCpgLibFeb09.pl Probe prep schema.
    • 69,263 were designed.
    • Divided into four subsets based on dG:
    • Set I: dG <= -20; 461 (19 are too long) => 1147 non-degenerate probes
    • Set II: -20 < dG <= -8; 15,871 (1212 are too long) => 24,322 non-degenerate probes
    • Set III: -8 < dG < 0; 53,033 (1892 are too long) => 70,294 non-degenerate probes (IIIa: top 15,294 probes; IIIb: bottom 55,000 probes)
    • Set IV: dG > 0; 1,122 (32 are too long)=> 1,313 non-degenerate probes
    • The total number of non-degenerated probes is 97,076, which capture 67,332 targets.
    • The top 55,000 oligos (-8<dG<-2) are in CpgProbes_Jan_55kA;
    • The remaining 42,076 oligos are in CpgProbes_Jan_55kB.
      • 9136 Exome-March-09 non-unique probes (amplified with the V4 primers) were also included in this library.
      • The remaining spaces were filled by all probes in Set I and some in Set II.
    • 103nt, MW=34kD

Agilent probes, V2 (April 2008)[edit]

  • Why redesigning probes?
    • V1 design was based on Primer3. Many constrains in Primer3 are probably either unnecessary or inappropriate for padlock probes.
    • We would like to search for an optimal (maximal coverage, minimal number of CpG in H1 & H2) set of probes covering a CpG island. It is hard to integrate Primer3 into the search of such an optimal set.
    • The capturing sequences should be as specific as possible. They don't have to be completely unique however, since a low percentage of non-specific capture is acceptable.
  • What are the differences in the V2 algorithm?Code.
  1. The codes for the search of capturing sequences were completely rewritten.
    1. The algorithm first search for all oligos with appropriate length and Tm, then look for pairs with appropriate gap size among all pair-wise combinations of these oligos.
    2. These pairs were assigned scores based on criteria including Tm, size, gap size, number of CpGs and a frequency score based on word statistics (word size K=12).
    3. Because the number of such pairs could be very large, the target region was then divided into a number of 10-bp bins, and only the pair with the best score in each bin was selected for the next round of optimization.
    4. These pairs are then assembled into sets. These sets cover every combination of the pairs. Each set has a score, which is based on the score of each pairs, plus the coverage of all pairs in the set.
    5. The candidate sets are then sorted based on the score, and the one with the best score was selected.
  2. To evaluate the specificity of capturing sequences, the counts of all 12-mers in both strands of the bisulfite converted genome were calculated, and saved as a data file. This file was loaded into the memory during probe design. The frequency score of each sequence is a weighted sum of all 12-mer sub-sequences. This calculation has a directionality, because the specificity for the sub-sequence close to the gap (for ligation or extension) should be higher. So the weight for the first 12-mer next to the gap is 1, and it is reduced by 25% per base as the sub-sequence moves outwards. I also added a hash table as a buffer so that the same oligo doesn't have to be calculated over and over again.

cpgMIP30k probe set[edit]

Two chromosomes enriched in genes differentially expressed between hES/iPS and fibroblasts.

                  # probes  total bps covered   CGI covered    %CGI covered
  Chromosome 12    15,998    973,377/924,553  799,818/767,129   82.1%/83.0%
  Chromosome 20    11,651    654,755/693,920  576,059/550,144   83.0%/84.0%

Additional space for other targets:

  1. Key genes in maintaining pluripotency or differentiation. 2000bp up/down-stream of the Transcription Start Sites (TSS, based on SwitchGear TSS), ES-Fib-diff: 780 probes
  2. Genes suggested by the Hochedlinger lab (MEG3,SNRPN,CDX1,HAND1,PAX6,FGF7) KH: 293 probes. Gene list of #1 & #2.
  3. ENCODE regions, Stanf promoters:
 ENm004: Chr22         151 probes    chr22:30,133,954-31,833,953
 ENm005: Chr21         271 probes    chr21:32,668,237-34,364,221
 ENm006: ChrX          307 probes    chrX:152,767,492-154,063,081
 ENm007: Chr19         276 probes    chr19:59,023,585-60,024,460
 ENm010: HOXA cluster  135 probes    chr7:26,924,046-27,424,045
 ENm011: 1GF2/H19      188 probes    chr11:1,699,992-2,306,039
 ENr123: Chr12          10 probes    chr12:38,626,477-39,126,476
 ENr333: Chr20         157 probes    chr20:33,304,929-33,804,928
  
  Added:
  ES-Fib-diff 776 (duplicate probes)
  Total      30,000
  Split into three sets: merge chr12 and chr20 probes, sorted by H1/H2 specificity
  cpgMIP30kV2.1: low score
  cpgMIP30kV2.2: medium score  
  cpgMIP30kV2.3: high score + other subsets
  Probe information
  Capturing Efficiencies based on end-sequencing.

Additional probes for normalizing the cpgMIP30k set[edit]

  • A cpgMIP30k captured library was end-sequenced on 06/12/08, 10364/10582 targets showed up at least once. So I sorted all 10582 probe templates in the order from low abundance to high (abundance for the ones did not showed up were considered 0). These templates were converted to probes with probe2padlockCpgLibV2.pl, and split into four sets (the last one not synthesized).
  1. cpgMIP30kNormL: 4677 probes, abundance: 0-10; the remaining space was filled with 5323 CES22k set on the low end of the distribution on gDNA capture (based on Hues62/63 data on 06/12/08).
  2. cpgMIP30kNormM: 9,999 probes (one included twice), abundance: 11-80
  3. cpgMIP30kNormH: 10,000 probes, abundance: 80-881

Agilent probes, V1[edit]

  • Several considerations in this design include:
    • Degenerate bases: I allowed at most two degenerate bases per primer. To convert PCR primers to padlock probes, multiple probes will be designed for one primer pair in order to cover all combinations.
    • Gap size: The range of amplicon size is 225-275bp, so the gap size is approximately 185-225bp.
    • Amplification adaptors: Version 4; Linker sequences: Version 7
   AP1: GTAGACTGGAAGAGCACTGTT  V4
   AP2: GATCGGATACGCATGAGGCTA  V4
   Linker: GTTGGAGGCTCATCGTTCCTATTCAGCTGCAGATGTTATCGAGGTCCGAC
           ------                  ----                ------
           Mme I                   Alu I                Mme I
   
  • Here is the perl code for primer design. The input file is cpgIslandExt.txt downloaded from UCSC GoldenPath database.
  • Conversion of PCR primers to padlock probes: AP1 + revcomp(FP) + linker + RP + AP2. codes.
  Some statistics
  Total probes:                     210,498
  Total CpG islands:                 23,496
  Total non-overlapping regions: 12,037,010 bps
  Total captured sequences:      13,291,608 bps
  ===CpgMIP28k set===
          probes                   CpG islands
  Chr6:   9,875          3,028    (1103/1251, 559kb/940kb)
  Chr20:  6,197          1,930    (695/779, 358kb/613kb)
  Chr21:  2,619            809
  Chr22:  5,358          1,664    (622/716, 310kb/562kb)
  Chr19:  first 3,857

Column synthesized probes (36)[edit]

  • Kun:LabNotes/CpgSeq/2007-10-12: probe = revcomp(RP) + linker + FP
  • Just realized that there was a mistake in probe design (11-20-2007). I assumed that after bisulfite treatment, the two strands are still complementary, which is not true. The probes have to be designed based on one strand only. If the forward PCR primers contain no "C" and the reverse primers contain no "G", that means the target strand is the forward strand. In that case, the padlock probes should be: revcomp(FP) + linker + RP. Here are the revised perl scripts: first 24 primers; second 12 primers. Here is the new probe file.

Read mapping[edit]

Library construction[edit]