Athurva Gore/LabNotes/2009-8-18: Difference between revisions

From ZhangLabWiki
Jump to navigation Jump to search
>Ajgore
>Ajgore
 
(9 intermediate revisions by the same user not shown)
Line 11: Line 11:
** LeeXGenes (on opposite strand to target cDNA)
** LeeXGenes (on opposite strand to target cDNA)
** ZhangSNP (on opposite strand to target cDNA)
** ZhangSNP (on opposite strand to target cDNA)
** FlyDup901
* '''CURRENTLY RUNNING:'''
* '''CURRENTLY RUNNING:'''
** FlyDup901
** FlyDup5279
** FlyDup5279
* '''TO RERUN:'''
* '''TO RERUN:'''
** TDMR Set (Also check if out-of-memory happens again...maybe run on Miner?)
** TDMR Set (Also check if out-of-memory happens again...maybe run on Miner?)
*** Look into this tomorrow
*** Look into this tomorrow
* Check numbering of new probes
** Appears to be slightly inaccurate as well.
** For residue coordinate system:
*** End position should be increased by 1
*** Start position should be increased by 2
** For space coordinate system:
*** End position should be increased by 1
*** Start position should be increased by 1


=Exome SNPs - NA12878=
=Exome SNPs - NA12878=
* Currently:
* Currently:
** Running script to separate our SNP calls and 1KG SNP calls by in-range and out-of-range
** Running script to separate our SNP calls and 1KG SNP calls by in-range and out-of-range
** NOTE:  File Dr. Zhang provided uses a slightly inaccurate start position; needs to be incremented by 1 for the residue based coordinate system.  End is accurate.
==Specific vs. Nonspecific capture==
* For our SNP calls:
* For our SNP calls:
** FOUND THAT 3190/20762 were NONSPECIFIC
** FOUND THAT 3190/20762 were NONSPECIFIC
** ~85% specific capture
** ~85% specific capture
* Want to separate false positives from true positives in our SNPs.
* Want to separate false positives from true positives in our SNPs.
* NEXT STEP:
==Comparison of our SNP calls to 1KG==
** Look at IPS data; plot quality and coverage of SNPs.
* Compared our SNP calls to 1KG
** Run bowtie on runs from NA12878 that have high false positives
** Found (out of 20,761 SNPs):
** See if there is a pattern; can we separate these false positives out somehow?
** 16319 hits (~78.5%)
** 598 miscalls (~3%)
** 3844 misses (~18.5%)
* Need to determine reasons for this
** Plot of quality and coverage:
 
==NEXT STEPS:==
* Look at IPS data; plot quality and coverage of SNPs.
* Run bowtie on runs from NA12878 that have high false positives - Done
* See if there is a pattern; can we separate these false positives out somehow?


=NA12878 High False Positive Reads=
=NA12878 High False Positive Reads=
Line 34: Line 53:
** 07282009_HL002, Lanes 4-7
** 07282009_HL002, Lanes 4-7
** 08052009_HL003, Lanes 4-5
** 08052009_HL003, Lanes 4-5
* Storing data on genome-miner:/media/TmpStore1/AG_Scratch/Exome/NA12878
* Using this sample for validation
* Using this sample for validation
** First step, need to map these reads.
** First step, need to map these reads.
** Start out with reads per experiment
** Start out with reads on a per experiment basis
*** Merge s_4 through s_7 in HL002, merge s_4 and s_5 in HL003
*** Merge s_4 through s_7 in HL002, merge s_4 and s_5 in HL003
*** Run Bowtie-SAM script on these merged sets; check results
*** Run Bowtie-SAM script on these merged sets; check results
*** Bowtie-SAM has been run on both HL003 and HL002.
* HL002 has a TON of SNPs.
** 185k!
** Something is very odd here.
* Noticed that HL002 and one set from HL003 both used size-selection.
** Not sure about Billy's results
* Looked for a homozygous overrepresentation.
** In HL002, definite presence of homozygous SNPs over heterozygous SNPs.
* Next Steps:
** Check both of these samples against dbSNP
** Check both of these samples against 1KG
** Is there anything similar among the false positives?

Latest revision as of 02:44, 19 August 2009

Navigation[edit]

Probe Generation[edit]

  • DONE
    • KKESH72
    • RPLCACRD
    • CpG-SNP set (with 80 bp gap)
    • FlyDup936
    • A-to-I Probes for Erez and Billy
    • LeeCancer (on opposite strand to target cDNA)
    • LeeXGenes (on opposite strand to target cDNA)
    • ZhangSNP (on opposite strand to target cDNA)
    • FlyDup901
  • CURRENTLY RUNNING:
    • FlyDup5279
  • TO RERUN:
    • TDMR Set (Also check if out-of-memory happens again...maybe run on Miner?)
      • Look into this tomorrow
  • Check numbering of new probes
    • Appears to be slightly inaccurate as well.
    • For residue coordinate system:
      • End position should be increased by 1
      • Start position should be increased by 2
    • For space coordinate system:
      • End position should be increased by 1
      • Start position should be increased by 1

Exome SNPs - NA12878[edit]

  • Currently:
    • Running script to separate our SNP calls and 1KG SNP calls by in-range and out-of-range
    • NOTE: File Dr. Zhang provided uses a slightly inaccurate start position; needs to be incremented by 1 for the residue based coordinate system. End is accurate.

Specific vs. Nonspecific capture[edit]

  • For our SNP calls:
    • FOUND THAT 3190/20762 were NONSPECIFIC
    • ~85% specific capture
  • Want to separate false positives from true positives in our SNPs.

Comparison of our SNP calls to 1KG[edit]

  • Compared our SNP calls to 1KG
    • Found (out of 20,761 SNPs):
    • 16319 hits (~78.5%)
    • 598 miscalls (~3%)
    • 3844 misses (~18.5%)
  • Need to determine reasons for this
    • Plot of quality and coverage:

NEXT STEPS:[edit]

  • Look at IPS data; plot quality and coverage of SNPs.
  • Run bowtie on runs from NA12878 that have high false positives - Done
  • See if there is a pattern; can we separate these false positives out somehow?

NA12878 High False Positive Reads[edit]

  • Locations
    • 07282009_HL002, Lanes 4-7
    • 08052009_HL003, Lanes 4-5
  • Storing data on genome-miner:/media/TmpStore1/AG_Scratch/Exome/NA12878
  • Using this sample for validation
    • First step, need to map these reads.
    • Start out with reads on a per experiment basis
      • Merge s_4 through s_7 in HL002, merge s_4 and s_5 in HL003
      • Run Bowtie-SAM script on these merged sets; check results
      • Bowtie-SAM has been run on both HL003 and HL002.
  • HL002 has a TON of SNPs.
    • 185k!
    • Something is very odd here.
  • Noticed that HL002 and one set from HL003 both used size-selection.
    • Not sure about Billy's results
  • Looked for a homozygous overrepresentation.
    • In HL002, definite presence of homozygous SNPs over heterozygous SNPs.
  • Next Steps:
    • Check both of these samples against dbSNP
    • Check both of these samples against 1KG
    • Is there anything similar among the false positives?