SEQanswers

Go Back   SEQanswers > Applications Forums > Sample Prep / Library Generation



Similar Threads
Thread Thread Starter Forum Replies Last Post
ChIP seq_why size selection pre PCR? EGraf Sample Prep / Library Generation 0 03-31-2015 01:06 AM
Ampure XP beads Dual Bead-based Size Selection ratio and PCR purification Leo Lee Sample Prep / Library Generation 4 09-19-2014 10:36 AM
What insert size for velveth (1.2.10) with 2 sets of reads with diff. insert sizes? Genomics101 Bioinformatics 4 02-07-2014 12:41 PM
Insert size != Fragment size? Boel Bioinformatics 6 12-12-2013 09:28 AM
Is the PCR purification step between ligation and size selection necessary? skblazer Sample Prep / Library Generation 7 10-28-2011 04:44 AM

Reply
 
Thread Tools
Old 07-12-2016, 06:35 AM   #1
melop
Member
 
Location: college station

Join Date: Jan 2012
Posts: 22
Default Always get the same insert size after PCR despite initial size selection on gDNA

Dear All, I am trying to make our own WGS library with dual barcodes. It involves a very common workflow including fragmentation (we use a Shearase from Zymo instead of covaris), End-Repair + A-tailing, Y-shape adapter ligation and PCR amplification.

After shearing, I cleaned up the gDNA fragments with 1:1 SPRI beads. A tapestation run shows that this shifts the peak distribution from 170bp to 271 bp. I then continued with the other steps. After PCR however, the fragment size turns out to be around 303 bp, minus the adapters it's around 160bp, despite the fact that I have done a size selection at the beginning.

Based on your experience, is this mainly due to PCR bias towards small fragments (we use KAPA Hifi mix)? I also included a negative control from the start of the workflow to make sure that these small fragments are not contaminants or some other random amplicons.
Attached Files
File Type: pdf cutband.pdf (134.5 KB, 28 views)
melop is offline   Reply With Quote
Old 07-12-2016, 05:38 PM   #2
adam.geber
Member
 
Location: United States

Join Date: May 2014
Posts: 40
Default

1) The D5000 assay has ±10% sizing accuracy, which is within the range of the fragment sizes you're seeing. Have you seen a size discrepancy for your libraries across multiple preps?

2) PCR bias towards smaller fragments is definitely a factor, although this can be partially mitigated by increasing your extension time. I'm assuming that you're using a KAPA prep kit -- have you looked in their technical guide for size distribution optimization? They have some good figures showing the the effects of different size selection methods and some recommendations that seem quite relevant to your question.

3) Finally, the libraries you show are post-PCR amplification and pre-cleanup, correct? Why not just do an additional SPRI-based size selection after your PCR enrichment?

Last edited by adam.geber; 07-12-2016 at 09:33 PM.
adam.geber is offline   Reply With Quote
Old 07-12-2016, 11:08 PM   #3
melop
Member
 
Location: college station

Join Date: Jan 2012
Posts: 22
Default

Dear Adam, thank you for the suggestions.
Quote:
Originally Posted by adam.geber View Post
1) The D5000 assay has ±10% sizing accuracy, which is within the range of the fragment sizes you're seeing. Have you seen a size discrepancy for your libraries across multiple preps?
I always get a peak at around 303bp after PCR, regardless of what size selection I tried before PCR. The first figure shows the fragmented DNA (without adapters), while the second one includes adapters (~150bp in my case). So the size discrepancy is not explained by the sizing precision of the D5000 assay. Now I'm wondering if the primers have non-specific bindings to the genomic DNA, so it always amplifies the same peak. I will try to use raw gDNA for a PCR to find out.

Quote:
Originally Posted by adam.geber View Post
2) PCR bias towards smaller fragments is definitely a factor, although this can be partially mitigated by increasing your extension time. I'm assuming that you're using a KAPA prep kit -- have you looked in their technical guide for size distribution optimization? They have some good figures showing the the effects of different size selection methods and some recommendations that seem quite relevant to your question.
We are not using the KAPA kit, but only the KAPA Hifi premix for the PCR step. I have read through the guide, doesn't seem to talk much about size distribution bias during PCR or the ways to mitigate the effect. It has some guides on bead size selection, which is something I'm already doing. It suggests that getting the fragment size distribution correct from the beginning is important since the bead size selection doesn't produce a tight range.

Quote:
Originally Posted by adam.geber View Post
3) Finally, the libraries you show are post-PCR amplification and pre-cleanup, correct? Why not just do an additional SPRI-based size selection after your PCR enrichment?
Yes I have tried the double cut SPRI selection. The post-PCR size selection isn't very helpful, it shifts the peak to the right by ~30bp (330bp), but due to the much higher peak in the small size range in the PCR product, it doesn't generate my desired the insert size (350-650bp fragment, 200-500bp insert). This is why I try to optimize the distribution after PCR to make sure that it's as close to the target size as possible.
melop is offline   Reply With Quote
Old 07-13-2016, 09:47 AM   #4
melop
Member
 
Location: college station

Join Date: Jan 2012
Posts: 22
Default

Today I performed more tests. I adjusted the fragmentation step to get a bigger gDNA distribution. I also left the beads in after ligation clean-up for PCR. This seems to help with the size distribution a bit. The longest fragment size I can get now is 372bp (insert 222 bp). However the sheared DNA should have a peak at 354bp, plus 150bp of adapter that should be around 500bp. So my final library is still 130bp shorter than expected...

One thing I notice is that when I ran the ligated fragments, the Y-shaped adapter makes them run slower. But there seems to a second, lower peak. Are these unligated fragments? If so perhaps I can increase adapter input (currently I'm using 1ul of 5uM adapter mix in 20ul reaction) to make sure they are ligated?
Alternatively, perhaps the End-repair and A-tailing wasn't efficient enough, so the longer fragments in the library don't get the adapter ligation?

Any pointers would be appreciated!
Attached Images
File Type: png sheared_beadselected.png (78.9 KB, 14 views)
File Type: png library_ligation.png (139.2 KB, 10 views)
melop is offline   Reply With Quote
Old 07-13-2016, 03:10 PM   #5
nucacidhunter
Jafar Jabbari
 
Location: Melbourne

Join Date: Jan 2013
Posts: 1,187
Default

Could you just post overlapped profile of sheared clean input and corresponding PCR after cleanup.

There is no reason for preferential ligation at this size range.
nucacidhunter is offline   Reply With Quote
Old 07-13-2016, 03:17 PM   #6
melop
Member
 
Location: college station

Join Date: Jan 2012
Posts: 22
Default

Quote:
Originally Posted by nucacidhunter View Post
Could you just post overlapped profile of sheared clean input and corresponding PCR after cleanup.

There is no reason for preferential ligation at this size range.
Haven't done the cleanup yet because in the protocol I am using it asks for a double cut bead selection, after which the distribution will differ from the raw pcr. If I use a 1.8x clean up the peak doesn't shift.
melop is offline   Reply With Quote
Old 07-13-2016, 04:43 PM   #7
nucacidhunter
Jafar Jabbari
 
Location: Melbourne

Join Date: Jan 2013
Posts: 1,187
Default

For these libraries double SPRI is not required but PCR need to be cleaned before sequencing anyway, so you can do 1x or 0.8x bead clean up (depending on the ratio used for clean up of fragmented input DNA). This will have the same size cut off as input and should not change the size distribution.
nucacidhunter is offline   Reply With Quote
Old 07-13-2016, 10:29 PM   #8
melop
Member
 
Location: college station

Join Date: Jan 2012
Posts: 22
Default

Quote:
Originally Posted by nucacidhunter View Post
For these libraries double SPRI is not required but PCR need to be cleaned before sequencing anyway, so you can do 1x or 0.8x bead clean up (depending on the ratio used for clean up of fragmented input DNA). This will have the same size cut off as input and should not change the size distribution.
Right, so a PCR clean up won't help much with shifting the distribution up (to peak at 450bp )?
melop is offline   Reply With Quote
Old 07-13-2016, 11:28 PM   #9
nucacidhunter
Jafar Jabbari
 
Location: Melbourne

Join Date: Jan 2013
Posts: 1,187
Default

DNA migration speed and hence calculated size is affected by presence of salts and protein in the sample. For fair comparison fragmented input DNA and PCR product has to be resuspended or eluted in the same buffer.

If you clean the PCR product with lower ratio of bead to amplicon (for instance 0.65x) short library fragments below 300bp will be cut off and that will shift the remaining library fragments peak to higher value. But I do not think that is what you want and is not advisable as it will reduce library diversity and yield.
nucacidhunter is offline   Reply With Quote
Old 07-13-2016, 11:49 PM   #10
melop
Member
 
Location: college station

Join Date: Jan 2012
Posts: 22
Default

Quote:
Originally Posted by nucacidhunter View Post
DNA migration speed and hence calculated size is affected by presence of salts and protein in the sample. For fair comparison fragmented input DNA and PCR product has to be resuspended or eluted in the same buffer.

If you clean the PCR product with lower ratio of bead to amplicon (for instance 0.65x) short library fragments below 300bp will be cut off and that will shift the remaining library fragments peak to higher value. But I do not think that is what you want and is not advisable as it will reduce library diversity and yield.
Thanks! I will try cleaning with 1X beads then to see if the size becomes expected.

Another question though is the double peaks after ligation. From previous posts some seems to suggests that the lower peak around 300bp represent fragments with a single adapter ligated, while the very "long" fragments are caused by Y-shaped adapters on both sides. Does this suggest insufficient adapter input or poor ligation in my case since the lower peak is quite high?
melop is offline   Reply With Quote
Old 07-14-2016, 12:41 AM   #11
nucacidhunter
Jafar Jabbari
 
Location: Melbourne

Join Date: Jan 2013
Posts: 1,187
Default

I have seen the double peak in TruSeq prepared libraries as well and since the kits normally are optimised I do not think that it is due to low quantity of adapters. The explanation about its origin is logical but I have not seen any experimental evidence to support that.

If the input is FFPE sample reduction in library size is normall as some larger fragments will not be amplified due to damage.

Last edited by nucacidhunter; 07-14-2016 at 12:44 AM.
nucacidhunter is offline   Reply With Quote
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off




All times are GMT -8. The time now is 12:06 AM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2018, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO