Kun:LabNotes/ExonomeSeq: Difference between revisions

From ZhangLabWiki
Jump to navigation Jump to search
mNo edit summary
Line 1: Line 1:
2010
<calendar>
name=Kun:LabNotes/ExonomeSeq
format=%name/%year-%month-%day
date=2010/02/01
view=oneyear
</calendar>
2009
2009
<calendar>
<calendar>

Revision as of 07:07, 11 January 2010

2010 <calendar> name=Kun:LabNotes/ExonomeSeq format=%name/%year-%month-%day date=2010/02/01 view=oneyear </calendar>

2009 <calendar> name=Kun:LabNotes/ExonomeSeq format=%name/%year-%month-%day date=2009/02/01 view=oneyear </calendar>

2008 <calendar> name=Kun:LabNotes/ExonomeSeq format=%name/%year-%month-%day date=2008/02/01 view=oneyear </calendar>

2007 <calendar> name=Kun:LabNotes/ExonomeSeq format=%name/%year-%month-%day date=2007/11/01 view=threemonths </calendar>

Probe Design

Some key numbers

  • RefExon: 207,671 (compiled from refGene.txt, hg18), total length: 60,879,002bp
  • Coding refExons: 166,554, length of coding seq: 29,118,349bp
  • Coding refExons + 6bp splicing motifs: 309,387,676bp
  • CCDS exons: 165,146; total size (incl. 6bp splicing motif) 30,141,138bp
    • CORRECTED: CCDS exons: 163,641; total size (incl. 6bp splicing motif) 29,860,098bp
  • miRNA genes: 685, 61,314bp

Dec08 whole exome probe set

  • In this round of probe design, constrains for A/C content and A/C homopolymers were included.
  • Two slightly different probe sets were designed.

Dec08 set: Probe file

First Round
  • First pass (code): Capturing arms do not contain repetitive sequences or SNPs (SNP129)
    • Number of probes: 305,124
    • Coverage: 83.5% (percentage of bases covered)
    • Efficiency: 77.5% (percentage of bases covered once)
  • Second pass (code): Allow repetitive sequences or SNPs outside of the terminal 12bp regions of the capturing arms:
    • Number of probes: 21,905
    • Combined coverage: 89.4%
    • Combined efficiency: 76.1%
    • Total number of probes that capture unique targets (<90% homology with other sequences): 304,295(RefSeq)/264,245(CCDS).
    • Number of targets that are mostly repeats (average 85% repetitive sequences) in the CCDS set: 456
    • Number of remaining probes: 263,789
    • Number of probes synthesized (H1+H2<=49bp): 258,861 (4,928 need longer oligos); 20065 CCDS genes, 148,649 coding refExons (25,648,790bp, target (incl. splicing motifs): 27,249,854bp; covered by the probes: 24,932,529bp)
    • Probes that contain DpnII site in H1/H2: 27,941 (DpnII SS set)
    • Probes that do not contain Dpn II site: 230,920 (DpnII DS set)
    • Region covered by two probes: 4,670,165bp
    • Region outside targets: 7,044,705bp (78% on-target)
    • Capturing arms for the 258,861 synthesized probes:RefExonJan09-all-sorted.txt(12.6MB)
    • These probes were remapped to the ccdsExon set generated by Athurva. It turned out that 1,595 probes were unmappable. These were the probes that target non-CCDS exons. ccdsExonJan09-all-sorted-unique.txt(12.6MB)
Second round (for missing ccdsExons)
    • Some exons were missing because they were on chrN_random based on the refGene annotation. Athurva has generated a list of ccdsExon and ccdsExonLink based on ccdsGenes downloaded from UCSC. I used this perl script to map all probes in the Jan09 set to ccdsExonLink, and generated a list of the missing ccdsExons (17,111 in total).
    • I used getExonProbe1stRound.pl to do the first pass of probe design on the list of missing ccdsExons. In this round of design, SNPs or repetitive sequences are not allowed in H1/H2.
    • Then I ran getExonProbe2ndRound.pl on the remaining 5578 ccdsExons. SNPs or repetitive seqeuences are allowed outside the 12bp terminal sequences of H1/H2.
    • For the 4020 exons that no probes could be designed in the previous two passes, I ran getExonProbe3rdRound.pl with even less stringent criteria:
      • Wider Tm range (48-75C);
      • Target length extended to 225bp;
      • Allow for more than one probe for small exons.
      • After this third design, only 93 exons were still missing.
    • A total of 33235 probes were designed for the missing ccdsExons after three rounds.
    • 27150/33235 probes capture unique targets (no homologous sequence with >=95% similarity)
    • Combining the 33,235 probes with the 258,861 probes ordered in the Jan09 set, a total of 27.3Mb (91%) CCDS exons were covered. 89% CCDS exons were covered by 27,150+258,861 probes.
    • [Feb-16-2009] Just realized that the ccdsExon set generated by Athurva contains 1,505 redundant exons(1679 probes), which were alternative splicing isoforms of RefSeq genes. After these redundant ones were removed (or the shorter ones combined to longer ones), the total CCDS set contains 163,641 unique exons. Including 685 miRNAs, we have a total of 164,326 targets. Some of the adjacent exons are separated by small introns of less than 200bp in size. Since we are including 80-120bp flanking sequences on both sides of an exon for probe design anyway, I decided to merge these adjacent exons if the intron size is less than 200bp. After the merging, I got 145,532 targets.
      • The list of missing exons was reduced from 17,111 to 15,744: missing unique ccdsExons
      • The number of probes for these unique exons was reduced to 22,271, among which 16,566 capture unique targets.
Third Round (Gap filling)
    • 89,479 missing exons or gaps were identified using this Perl script.
    • 37,487 probes were designed to cover most gaps, among which 33,953 capture unique targets.
    • The combination of 258,861 probes in Jan09 set, 22,271 probes in the 2nd round, and 37,487 probes in the 3rd round covers 98.5% of ccdsExons (29,424,482bp/29,860,098bp).
March-09 order
  • All round2/round3 probes capturing non-unique targets were placed into one set. Should be amplified with the V4 primers. Included in the CpgProbes_Jan_55kB order.
  • Round2 and Round3 unique probes are synthesized in one 55k library. Round2 probes (16,566) should be amplified with the V4 primers, and Round3 probes (33,949) with V6 primers. This library is 145-mers. Probes longer than 145 mers were trimmed from the H2 arm.
  • Included 685 probes for miRNA genes.
  • The remaining space of 3800 probes was filled with Round2 probes with lowest folding energy.
Read mapping
  • Mapping reads to the captured targets instead of the full genome will increase the speed by ~100x. However there are several considerations and practical issues in using this strategy:
    • Captured targets are partially overlapping, so if we use the target sequences as the mapping templates, some reads will have more than one identical hits.
    • Some flanking intronic sequences were also captured because the way the probes were designed. These sequences needs to be included as mapping templates.
    • The target sequences are a large number of relatively short sequences. Existing read mapping and variant calling software (Maq, SOAP etc) were not designed for such templates. Ideally, we should merge all the mapping templates into a single file.
  • Here are the procedures I used to prepare a mapping template file:
    • Combine CCDS_exons_UCSD_HARVARD.BED file with miRna.txt file, giving rise to a complete list of CCDS exons plus all microRNA genes.
    • Sort the list based on chromosome positions, add 100bp upstream and downstream sequences to each exon, and merge overlapping exons (145,532 in total).
    • Retrieve the DNA sequence of these regions from the hg18 unmasked sequences using GrabExons.
    • Merge all the DNA sequences in a single fasta file. Adjacent sequences are separated by Ns such that each sequence start at 1, 501, 1001... etc. The step size is 500. A table containing the sequence IDs and the starting positions was generated, so that we can remap the variant back to the genome coordinate after variant calling.

28Dec08 set: Not used because there isn't a significant improvement in coverage.

      • First round (code): Allow repetitive sequences or SNPs outside of the terminal 12bp regions of the capturing arms:
        • Number of probes: 311,741
        • Coverage: 85.0% (percentage of bases covered)
        • Efficiency: 77.2% (percentage of bases covered once)
    • Second round:
        • Number of probes:18,439
        • Combined coverage: 89.9%
        • Combined efficiency: 76.0%

Oct08 whole exome probe set

  • Athurva created a list of non-redundant coding exons for all RefSeq genes.
  • Athurva has converted my Cpg probe design program for exon probe design. He has made a number of changes including the strategy to search for a optimal set of probes for a large exon. I made additional changes on the codes. Here is the current version of probe design program.
  • A 27k probe set(exon102508) was ordered on 10-25-2008. This probe set contains four subset of oligos for different designs/projects.
    • Mike Chou ChrX probes(Len >=100bp, TypeIIs): 5904
    • Kun Chr21/22 probes (Nicking Enzymes): 2924+5654
    • Kun Chr20 probes (TypeIIs): 7334
    • MCC_probes(Celso/Feng, Bing Ren's lab): 5184
  • Two Agilent probe designs that allow for long linkers were tested:
    • TypeIIs design:[Codes] this is similar to the previous designs in that AP1V4/AP2V4 sequences that contain TspRI/DpnII Type IIs cutting sites were used.
    • Nicking Enzymes:[Codes] in this design, AP1V4/AP2V4 were replaced with eMIP_CA1_F/eMIP_CA1_R used by Jay Shendure in his eMIP 55k probe design. eMIP_CA1_F/eMIP_CA1_R contain recognition sites for nicking enzymes Nt.AlwI and Nb.BsrDl. Using nicking enzymes will simplified the probe prep step, also the digestion could be more complete and hence the yield could be higher (yet to be validated). The reason I stayed away from the method previously is that a small fraction of probes that contain the cutting sites within H1/H2 are not compatible with this method. My current thinking is that if it does turn out to be simpler and more efficient, we can make the majority of probe using this method, and the incompatible probes with the DpnII/USER method.
  • Received the oligo library on 11/20/2008
  • Length of some sequences:
    • Average captured fragment: 160bp
    • Capturing arms: 49bp
    • Sequencing adaptors/primers: 115bp
    • Average total length: 324bp = 214KDa.

The first probe set: Kun:LabNotes/ExonomeSeq/2007-11-15

  • One 55k library received on 12/12/2007, OLID: 018274110001, 10pmole, resuspended in 500ul RT-PCR grade water (new vial) -> 20nM x 500ul -> Take 100ul out to a 1.5ml non-stick tube as the working tube.
  • The second probe set: Kun:LabNotes/ExonomeSeq/2007-12-21
  • Two Agilent 55k libraries were ordered on 01/04/2008.
  1. Cosmic3760 0189191
  2. CpgCes: Cpg28k + CES27k 0189181

Received 01/24/2008. 10pmole, resuspended in 500ul RT-PCR grade water (new vial) -> 20nM x 500ul -> Take 100ul out to a 1.5ml non-stick tube as the working tube.

Overview of experiments

Sequencing library construction

Random primed

  • Forward primer (SolFN6):CGACGCTCTTCCGATCTNNNNNN Tm left = 59.20C
  • Reverse primer (SolRN6):CATACGAGCTCTTCCGATCTNNNNNN Tm left = 59.45C

USER-Lig

Feb09 protocol