Ylaine/2009-7-17: Difference between revisions

From ZhangLabWiki
Jump to navigation Jump to search
>Ylaine
(New page: ==Bwa/Sam filtered variants== * Re-do heatmap with consensus quality instead of SNP quality (see yesterday's entry) ==Maq/Sam== * Make heatmap ==SNP Filtering== * Look into reducing fals...)
 
>Ylaine
 
(9 intermediate revisions by the same user not shown)
Line 1: Line 1:
==Bwa/Sam filtered variants==
==Bwa/Sam filtered variants==
* Re-do heatmap with consensus quality instead of SNP quality (see yesterday's entry)
* Re-do heatmap with consensus quality instead of SNP quality (see yesterday's entry)
bwaSam/NA12878_40bp_061109.sequence.pileup.variants.txt
[[Image:Bwasam_dbSNP_q.jpg‎ ]]


==Maq/Sam==
==Maq/Sam==
* Make heatmap
* Make heatmap from:
MaqSamTarget/snp.nonunique.dbSNP
* Some SNPs in dbSNP have extremely high coverage, e.g. the highest coverage is at:
chr6    160384279              G      A/G    228    228    57      682    rs894817        A/G
** Set heatmap upper bounds to go to next highest coverage value, 396
[[Image:Maqsam_nu_dbSNP.jpg‎ ]]
* Try using duplicate-filtered SNP set for fewer funny numbers
awk '{print $1 "\t" $2 "\t" $3 "\t" $4 "\t" $5 "\t" $8}' snp.chr.txt > snp.chr.qr.txt
../Scripts/matchCnsSnp2Ref_dbSNP.pl snp.chr.qr.txt > snp.chr.dbSNP.txt
* False positives:
[[Image:Maqsam_dbSNP.jpg‎ ]]
grep 'rs' snp.chr.dbSNP.txt | ../Scripts/matchCnsSnp2Ref_AG.pl>snp.compare.in_dbSNP.txt
grep -v 'rs' snp.chr.dbSNP.txt | ../Scripts/matchCnsSnp2Ref_AG.pl>snp.compare.no_dbSNP.txt
{| {{table}}
| align="center" style="background:#f0f0f0;"|'''1K Genome'''
| align="center" style="background:#f0f0f0;"|'''In dbSNP'''
| align="center" style="background:#f0f0f0;"|'''Percent'''
| align="center" style="background:#f0f0f0;"|'''Not in dbSNP'''
| align="center" style="background:#f0f0f0;"|'''Percent'''
|-
| Match||7886||76.73%||216||1.69%
|-
| Miss||1137||11.06%||12506||97.66%
|-
| Mismatch||1255||12.21%||83||0.65%
|-
| Total||10277||||12805||
|-
|
|}
* FP rate: '''48.04%'''
[[Image:Maqsam_nodbSNP_rmdup.jpg‎ ]]


==SNP Filtering==
==SNP Filtering==
Line 11: Line 44:


==Other==
==Other==
* Re-calculate false positive percentage: divide by total number of SNPs, including those at dbSNP locations
* Summarize FP and FN rates for each method

Latest revision as of 22:09, 17 July 2009

Bwa/Sam filtered variants[edit]

  • Re-do heatmap with consensus quality instead of SNP quality (see yesterday's entry)
bwaSam/NA12878_40bp_061109.sequence.pileup.variants.txt

File:Bwasam dbSNP q.jpg

Maq/Sam[edit]

  • Make heatmap from:
MaqSamTarget/snp.nonunique.dbSNP
  • Some SNPs in dbSNP have extremely high coverage, e.g. the highest coverage is at:
chr6    160384279               G       A/G     228     228     57      682     rs894817        A/G
    • Set heatmap upper bounds to go to next highest coverage value, 396

File:Maqsam nu dbSNP.jpg

  • Try using duplicate-filtered SNP set for fewer funny numbers
awk '{print $1 "\t" $2 "\t" $3 "\t" $4 "\t" $5 "\t" $8}' snp.chr.txt > snp.chr.qr.txt
../Scripts/matchCnsSnp2Ref_dbSNP.pl snp.chr.qr.txt > snp.chr.dbSNP.txt
  • False positives:

File:Maqsam dbSNP.jpg

grep 'rs' snp.chr.dbSNP.txt | ../Scripts/matchCnsSnp2Ref_AG.pl>snp.compare.in_dbSNP.txt
grep -v 'rs' snp.chr.dbSNP.txt | ../Scripts/matchCnsSnp2Ref_AG.pl>snp.compare.no_dbSNP.txt
1K Genome In dbSNP Percent Not in dbSNP Percent
Match 7886 76.73% 216 1.69%
Miss 1137 11.06% 12506 97.66%
Mismatch 1255 12.21% 83 0.65%
Total 10277 12805
  • FP rate: 48.04%

File:Maqsam nodbSNP rmdup.jpg

SNP Filtering[edit]

  • Look into reducing false positives (not in dbSNP, not found by 1000 genome)
    • Pattern classification: generate PDF in Matlab using 'ksdensity'
    • Matlab's optimization toolbox

Other[edit]

  • Summarize FP and FN rates for each method