Dinh 2011/NOTES/2011-11-17

From ZhangLabWiki
Jump to navigation Jump to search

MUIR samples from Yang Xu's lab[edit]

  • 2HFF_P6 is progenitor for 2HFF8d_10
  • 218HFF_P7 is progenitor for 218HFF411 and 218HFF211
    • Note that: 218HHF* is how the files are originally file names, I changed the HHF to HFF for the end analysis
  • Number of CpG captured in each sample (no CG-SNP):
  418477 218HFF_P7.txt.cg-pos.BED.txt
  375514 218HFF211.cg-pos.BED.txt
  351674 218HFF411.cg-pos.BED.txt
  432937 2HFF8d_10.cg-pos.BED.txt
  504668 2HFF_P6.txt.cg-pos.BED.txt
  404638 Hues1.txt.cg-pos.BED.txt
  309784 Hues3.txt.cg-pos.BED.txt
  458459 Hues8.txt.cg-pos.BED.txt
  • Number of CG-SNP in each sample:
 13035 218HFF_P7.txt.cSNP-pos.BED.txt
 11524 218HFF211.cSNP-pos.BED.txt
 10799 218HFF411.cSNP-pos.BED.txt
 13146 2HFF8d_10.cSNP-pos.BED.txt
 15490 2HFF_P6.txt.cSNP-pos.BED.txt
 12433 Hues1.txt.cSNP-pos.BED.txt
  9614 Hues3.txt.cSNP-pos.BED.txt
 14041 Hues8.txt.cSNP-pos.BED.txt

Memory/Mutation[edit]

  • First to use getCpG*11082011.pl, need a set of allowable sites!
awk '{print $1"\t"$3;}' Hues*.BED.txt | sort -u > H1H3H8_allsites
wc -l H1H3H8_allsites 502459
  • Compare Hues1vsHues3, Hues1vsHues8, Hues3vsHues8:
 Test, #comparable sites, #different sites
 1vs3  233,817  2130/233817 = 0.9%
 1vs8  307,055  3619/307055 = 1.2%
 3vs8  243,655  1657/243655 = 0.7
  • Find variable sites:
 less *dmTable | awk '{if($2==1) print $0;}' | sed 's/:/\t/g' | awk '{print $1"\t"$2}' | sort -u > variableH1H3H8
 wc -l variableH1H3H8  4862
  • Reform subset sites list:
 cat variableH1H3H8 variableH1H3H8 H1H3H8_allsites | sort | uniq -u > H1H3H8_filteredsites
 wc -l H1H3H8_filteredsites 497597
  • Run GO.calcMEMMU.sh
  1. Set up mf tables such that first test compare to ES and second test compare to SOM
for f in mf_*.txt
do
       ./getCpgChsq+DiffMatrix_DD_*.pl $f 10 0.2 0.001 H1H3H8_filteredsites > $f.dmTable
       awk ' 
               function abs(value) { return (value<0?-value:value) };
               { if($2==1 && $3==1 && abs($4-$5) >= 0.2 && ( ($4<0 && $5>=0) || ($4>0 && $5<=0) )) print $0,"\tMEMORY_INCOMPLETE";
                 if($2==1 && $3==0 && abs($4-$5) >= 0.2 && abs($5) < 0.2 && abs($4) > abs($5) ) print $0, "\tMEMORY_COMPLETE";
                 if($2==1 && $3==1 && ( ($4<0 && $5<0) || ($4>0 && $5>0) )) print $0, "\tMUTATION";
                 if($2==0 && $3==1 && abs($4)<0.2) print $0,"\tPLURIPOTENCY";
               }
       ' $f.dmTable > $f.memmuTable
done;

Results[edit]

 # total sites tested
 212239 mf_218HFF211.txt.dmTable (1350 memory, 530 mutations)
 193521 mf_218HFF411.txt.dmTable (2162 memory, 586 mutations)
 244023 mf_2HFF8d_10.txt.dmTable (2144 memory, 1101 mutations)

Genome browser shots[edit]

Each loci was picked randomly from a list of memory or mutation sites. File:Muir-MemoryMutation-TCERG1L.png File:Muir-MemoryMutation-IRS2.png File:Muir-MemoryMutation-HOX.png

Scatter plots[edit]

  • Distances are iPS - ES or iPS - SOM
  • It looks like more memory sites are due to difficulty getting methylated - iPS is less methylated than ES.
  • And more mutation sites are due to sites getting methylated when it should not be.

File:Muir-MemoryMutation-scatterplot.png