Dinh/Dinh 2013/NOTES/2013-10-7: Difference between revisions

From ZhangLabWiki
Jump to navigation Jump to search
>Dinh
mNo edit summary
>Dinh
mNo edit summary
Line 1: Line 1:
= Blueprint Capture, low mapping issue =
= Blueprint Capture, low mapping issue =
* Low mapping can be caused by any of the following problems
* Low mapping can be caused by any of the following problems
# Contamination - generally, contamination would not be amplified by the last PCR step or sequenced due to lack of adapter sequences
# Contamination - contamination of library after capture would not be amplified by the last PCR step or sequenced due to lack of adapter sequences. Contamination of reagents during capture may lead to low mapping rates. Contamination may explain why we have such variable mapping rates, since each DNA sample had varying quality, it's possible that the lower quality samples would have more capture from contaminating sequences and thus lower mapping rate.
# Chimeric capture sequences - when imperfect hybridization of capture arms to genomic sequences which occurs frequently in the genome, leading to off-target & chimeric sequences. However, when we remove the capture arms, we should get recover the alignments for these regions, so chimeric capture sequences may not be the case.
##Do the amount of input converted gDNA correlate with mapping rate?
# Error from sequencing - Sequencing quality is good from HiSeq Rapid Run.  
##Do the captured sequences map to other genomes in NT database?
# Error from oligo synthesis - error from oligo synthesis would still allow us to recover the alignments from the off-target regions, & even so, should lower the specificity of capture, but this is not the case.  
# Chimeric capture sequences - When imperfect hybridization of capture arms to genomic sequences which occurs frequently in the genome, leading to off-target & chimeric sequences. Or when the capture arms capture each other.
##How many of the sequences map to the adapters, linkers, and primers?
##Do the sequences map to human repeats? - No, when mapping with Bowtie, most of the reads either map uniquely or not at all.
##Do the sequences map to non-human repeats? - Check by performing Blast to RepBase18.08
# Error from sequencing - We did not use a PhiX control for the MiSeq runs, but had a control for the HiSeq RapidRun (control was ~ 89% of flowcell, BP libraries was ~ 11%). The mapping rate of the HiSeq RapidRun was still low for many samples. Also, the majority of mapped reads have low error rates at the 3' end.  
# Error from oligo synthesis - error from oligo synthesis would still allow us to recover the alignments from the off-target regions. Even so, we should get lower the specificity of capture, but this is not the case.
== MEGAN analysis of unmapped reads==
== MEGAN analysis of unmapped reads==



Revision as of 20:51, 8 October 2013

Blueprint Capture, low mapping issue

  • Low mapping can be caused by any of the following problems
  1. Contamination - contamination of library after capture would not be amplified by the last PCR step or sequenced due to lack of adapter sequences. Contamination of reagents during capture may lead to low mapping rates. Contamination may explain why we have such variable mapping rates, since each DNA sample had varying quality, it's possible that the lower quality samples would have more capture from contaminating sequences and thus lower mapping rate.
    1. Do the amount of input converted gDNA correlate with mapping rate?
    2. Do the captured sequences map to other genomes in NT database?
  2. Chimeric capture sequences - When imperfect hybridization of capture arms to genomic sequences which occurs frequently in the genome, leading to off-target & chimeric sequences. Or when the capture arms capture each other.
    1. How many of the sequences map to the adapters, linkers, and primers?
    2. Do the sequences map to human repeats? - No, when mapping with Bowtie, most of the reads either map uniquely or not at all.
    3. Do the sequences map to non-human repeats? - Check by performing Blast to RepBase18.08
  3. Error from sequencing - We did not use a PhiX control for the MiSeq runs, but had a control for the HiSeq RapidRun (control was ~ 89% of flowcell, BP libraries was ~ 11%). The mapping rate of the HiSeq RapidRun was still low for many samples. Also, the majority of mapped reads have low error rates at the 3' end.
  4. Error from oligo synthesis - error from oligo synthesis would still allow us to recover the alignments from the off-target regions. Even so, we should get lower the specificity of capture, but this is not the case.

MEGAN analysis of unmapped reads

  • I mapped the first 1000 unmappable reads with blastn
 $blastn -p blastn -d nt -K 100 -e 100 -g F -F "m D" -a 16 -W 7 -i Indx12.fa -o Indx12.bwa.R1.blast
  • Result from MEGAN (Indx12, Indx13, Indx14 only):
 File:Indx12.bwa.R1.megan.png
 File:Indx13.bwa.R1.megan.png
 File:Indx14.bwa.R1.megan.png