Brandon:LabNotes/Project1/2012-7-20

From ZhangLabWiki
Jump to navigation Jump to search

Update to 2012/6/29, data analysis after sequencing[edit]

  • Libraries look like they were successfully created after cloning/transformation etc. Now continuing data analysis after the sequencing run

bioinformatic analysis[edit]

  • samples
sample # description barcode file
1 6 ng pure DNA Index 49 s_5_1_Indx49.unique.bed
2 600 pg pure DNA Index 50 s_5_1_Indx50.unique.bed
3 60 pg pure DNA Index 51 s_5_1_Indx51.unique.bed
4 10000 Cells Index 52 s_5_1_Indx52.unique.bed
5 1000 Cells Index 53 s_5_1_Indx53.unique.bed
6 100 Cells Index 54 s_5_1_Indx54.unique.bed


scripts run[edit]

  • perl script for mapping etc.
#!/usr/bin/perl -w
use strict;

my $in_file = $ARGV[0];

my $split = $in_file;
$split =~ s/.txt//;  #substitution

my $unmappable = $split . ".unmappable.txt";
my $samfile = $split . ".bowtie.sam";
my $bamfile = $split . ".bam";
my $sortedBam = $split . ".sorted";
my $sortedBam2 = $split . ".sorted.bam";
my $uniqueBam = $split . ".unique.bam";
my $bed = $split . ".unique.bed";
 
#print "\n$split\n";
#print "$unmappable\n";
#print "$samfile\n\n";

my $cmd = "bowtie -n1 -k1 --best --chunkmbs 10240 --strata -l32 -m1 -p4 --nomaqround --solexa1.3-quals --sam --un $unmappable /GenomeDB/HsGenome/hg19/hg19 $in_file $samfile";
print "\n$cmd\n";
system($cmd);

$cmd = "samtools view -S -b -o $bamfile $samfile";
print "\n$cmd\n";
system($cmd);

$cmd = "samtools sort $bamfile $sortedBam";
print "\n$cmd\n";
system($cmd);

#$cmd = "samtools index Index1_sorted.bam";
#print "$cmd\n";
#system($cmd);

$cmd = "samtools rmdup -s $sortedBam2 $uniqueBam";
print "\n$cmd\n";
system($cmd);

#$cmd = "genomeCoverageBed -bg -ibam $uniqueBam -g /home/bsos/software/BEDTools-Version-2.12.0/genomes/human.hg19.genome > $bed";
#print "\n$cmd\n\n";
#system($cmd);


  • shell script for multiple samples
for f in s_5_1_Indx*.txt
do 
       perl /media/Ext12T/BS_Scratch2/120712_HL125/perltest2.pl $f
done;

 
  • parameters used for generate .bed files for genome mapping
genomeCoverageBed -bg -ibam s_5_1_Indx49.unique.bam -g /home/bsos/software/BEDTools-Version-2.12.0/genomes/human.hg19.genome > s_5_1_Indx49.unique.bed




  • library complexity, clonal reads, and unique reads statistics for tagmentation/IVT generated library.
' sample total reads one alignment one alignment % failed to map fail map % supressed supressed % clonal reads clonal read% unique reads
Index49 6 ng pure DNA 8,153,450 2,988,691 36.66% 4,782,904 58.66% 370,803 4.55% 63,703 2.131% 2,924,988
Index50 600 pure pg 3,438,219 396,617 11.54% 2,988,275 86.91% 53,327 1.55% 4,596 1.159% 392,021
Index51 60 pure pg 2,457,611 21,791 0.89% 2,432,690 98.99% 3,130 0.13% 102 0.468% 21,689
Index52 10000 cells 3,691,004 1,139,024 30.86% 2,418,176 65.52% 133,804 3.63% 22,580 1.982% 1,116,444
Index53 1000 cells 6,941,784 1,619,425 23.33% 5,129,288 73.89% 193,071 2.78% 67,628 4.176% 1,551,797
Index54 100 cells 4,899,774 353,895 7.22% 4,498,074 91.80% 47,805 0.98% 21,793 6.158% 332,102
index53 1k cells downsampled 3,854,691 899,400 23.33% 2,847,985 73.88% 107,306 2.78% 28,016 3.115% 871,384
UCSCRep1 69,585,103 46,179,576 66.36% 2,258,954 3.25% 21,146,573 30.39% 22,721,422 49.202% 23,458,154
UCSCRep3 144,500,455 90,721,068 62.78% 15,251,224 10.55% 38,528,163 26.66% 28,285,232 31.178% 62,435,836
  • file size of .bed files
sample ' unique reads file size file '
6 ng pure DNA 2,924,988 79M s_5_1_Indx49.unique.bed
600 pure pg 392,021 9.9M s_5_1_Indx50.unique.bed
60 pure pg 21,689 551K s_5_1_Indx51.unique.bed
10000 cells 1,116,444 29M s_5_1_Indx52.unique.bed
1000 cells 1,551,797 41M s_5_1_Indx53.unique.bed
100 cells 332,102 8.3M s_5_1_Indx54.unique.bed



Sample total reads uniquely mapped unique map % failed to map fail map % not uniquely mapped not uniquly mapped % clonal reads clonal read % unique reads
gentle-lysis-500-cell 13,030,577 7,999,292 61.39% 2876464 22.07% 2,154,821 16.54% 7,136,718 89.22% 862,574
gentle-lysis-200-cell 4,720,341 2,722,830 57.68% 1181536 25.03% 815,975 17.29% 2,432,561 89.34% 290,269
gentle-lysis-100-cell 4,017,439 1,911,153 47.57% 1458224 36.30% 648,062 16.13% 1,710,288 89.49% 200,865
lyse+protease-500-cell 5,220,947 3,715,710 71.17% 520931 9.98% 984,306 18.85% 207,382 5.58% 3,508,328
lyse+protease-200-cell 4,492,276 3,077,980 68.52% 534772 11.90% 879,524 19.58% 165,647 5.38% 2,912,333
lyse+protease-100-cell 3,323,185 2,154,267 64.83% 416699 12.54% 752,219 22.64% 159,548 7.41% 1,994,719


  • Comparison between data from when tagmentation was performed that added orange and blue ends, versus my method using amplifcation with IVT.


    • comparing IVT method versus tagmentation only, more reads were obtained from tagmentation only data, but 89% were clonal reads. with IVT amplification method, less clonal reads obtained, but most were unmappable due to inefficiencies during the procedure. Still more reads were obtained with the 100 cell sample using IVT than either the 100 or 200 cell sample using only tagmentation.


    • looking at the the IVT method, many more unique reads were obtained in the 1000 cell sample than in pure DNA, though could be possible that there were more cells used when doing the DNA accessibility assay, versus comparable amount of DNA from the pure DNA sample.


    • in IVT method massive amount of unmappable reads. as [DNA] and cell # decreases. Though still got more reads for DNA accessibility samples that was achieved through tagmentation only.


    • in IVT method, alot less reads suppressed (repeats etc), but correlated with amount of reads that failed to map so isn't an issue. similar in both pure DNA and cell samples.


UCSC genome browser validation, comparision with tagmentation data[edit]

  • seems to be working just as well as when only tagmentation was performed on the samples.
  • cases where tagementation and IVT does not work as well as when only tagmentation was done, and cases where it works better than when only tagmentation was done.
  • also hard to tell since when comparing to only tagmentation different cell lines were used. Pure DNA sample is also a IPS cell line so harder to make comparisons.


  • chromosome 1, 91 Mb, good match
  • samples prefixed with old are from sample where only tagmentation only was performed.
  • samples with 100*cell_cc and 6* pure DNA are the new samples with IVT amplification.
    • can see it works as expected with pure DNA having less tags than accessible DNA for the same region.

File:Chr1 91 tagmentation versus IVT 100 cells comparison.png



  • chromosoem 7, 61 mb.
    • read counts are higher in controls though.

File:Chr7 61 tagmentation versus IVT 100 cells comparison.png



  • chromosome 21, 9.8 mb
    • in IVT amplified sample definetly more accessible than control
    • not the case with older tagmentation only samples.

File:Chr21 9.8 tagementation versus IVT 100 cells comparison good alignment.png

MACS analysis[edit]

  • looked at analysis of peak calling programs (for CHIP-seq) again and MACS is best at calling more peaks with datasets that have fewer reads.

MACS

macs14 -t s_5_1_Indx54.unique.bam -c s_5_1_Indx50.unique.bam -n 100cell_with_control600pg
  • after running MACS program, alot of the FDRs were messed up, with them being 100, probably since there are not enough control reads in the control sample used.


  • Not sure why 10,000 cell sample didnt work as well.
  • when comparing old 500 cell tagmentation sample to 1000 cell tagmentation then IVT sample it seems more efficient. More RNA was generated in the 1000 cell sample too.
sample total reads acc unique rds acc unique rds con total peak calls bp called sig minux X minus X
IVT 100 cell, 6 ng control 4,899,774 332,102 2,924,988 197 101,875 48,909 minus 4 count tags
IVT 100 cell, 600 pg control 4,899,774 332,102 392,021 135 66,910 31,073 minus 4 count tags
IVT 100 cell, 60 pg control 4,899,774 332,102 21,689 152 74,489 29,674 minus 4 count tags
old 100 cell 4,017,439 200,865 1,994,719 1579 637,575 143,891 minus 3 count tags
old 200 cell 4,720,341 290,269 2,912,333 477 248,256 104,743 minus 4 count tags
old 500 cell 13,030,577 862,574 3,508,328 901 539,451 307,184 minus 5 count tags
IVT 1000 cell dwnsmpled, 6 ng con 3,854,691 871,384 2,924,988 1179 644,539 354,740 minus 5 count tags
IVT 1000 cell, 6 ng control 6,941,784 1,551,797 2,924,988 5339 3,070,305 1,910,345 minus 5 count tags
IVT 1000 cell, 600 pg control 6,941,784 1,551,797 392,021 4381 2,521,586 1,576,574 minus 5 count tags
IVT 1000 cell, 60 pg control 6,941,784 1,551,797 21,689 5325 3,040,347 1,808,772 minus 5 count tags
IVT 10000 cell, 6 ng control 3,691,004 1,116,444 2,924,988 657 364,017 158,185 minus 5 count tags


  • to directly compare 1000 cell sample to the 500 cell old tagmentation sample, the 1000 cell sample was downsampled so the same amount of unique reads were obtained after analysis. The original demultiplexed text file was downsampled, and then run through the same computational pipeline as the other files in the original analysis
  • Can see even though a similar amount of unique reads were obtained in the 500 cell sample and in the 1000 cell IVT downsampled sample, more peaks were called and bp covered in the genome. That is with 9.1 million fewer total reads in the IVT sample.


sample total reads acc unique rds acc unique rds con total peak calls bp called sig minux X minus X
old 500 cell 13,030,577 862,574 3,508,328 901 539,451 307,184 minus 5 count tags
IVT 1000 cell dwnsmpled, 6 ng con 3,854,691 871,384 2,924,988 1179 644,539 354,740 minus 5 count tags




genome coverage to see reporducability

coverageBed -a 1000cell_with_control600pg_peaks.bed -b 1000cell_with_control60pg_peaks.bed > 1000cell600pg_onto_1000cell60pg_coverage.txt
sample bases covered total length % covered '
100 cell 600 pg onto 100 cell 60 pg 60,084 74,489 0.806615742
100 cell 60 pg onto 100 cell 600 pg 60,084 66,910 0.897982364 Thus less peaks called in 100 cell 600 pg samples (makes sense)
100cell60pg_onto_1000cell6ng_coverage.txt 31,365 3,070,305 0.010215597
1000cell6ng_onto_100cell60pg_coverage.txt 31,365 74,489 0.421068883 expect more of smaller smaple to be covered by larger sample?
100cell600pg_onto_1000cell6ng_coverage.txt 31,383 3,070,305 0.01022146
1000cell6ng_onto_100cell600pg_coverage 31,383 66,910 0.469033029 expect the same here.
1000cell600pg_onto_1000cell60pg 2,313,603 3,040,347 0.760966758 24ish% loss in reprodability when using different control
1000cell600pg_onto_1000cell6ng 2,371,737 3,070,305 0.772476024 24ish% loss in reprodability when using different control
1000cell6ng_onto_10000cell6ng 58262 364017 0.160052965 same type of coverage stats as in tagmentation only
10000cell6ng_onto_1000cell6ng 58262 644539 0.090393289 same type of coverage stats as in tagmentation only
see comparison after MACS analysis for old data
used control 500 cell and control 100 cell for input into macs
500cell_onto_100cell_coverage 124,211 878,558 0.141380535 same type of coverage as in tagmentation IVT samples
100cell_onto_500cell_coverage 85,136 539,451 0.157819709 same type of coverage as in tagmentation IVT samples

Future[edit]

  • repeat procedure but use DNase after IVT, should hopefully solve alot of the unmappable reads issues.


  • instead of using pure DNA as the control, just use protease treated cells from the same sample. Then won't have the issue of possible differences in the amount of DNA tagmentation is being performed on.