SEQanswers

Go Back   SEQanswers > Sequencing Technologies/Companies > Illumina/Solexa



Similar Threads
Thread Thread Starter Forum Replies Last Post
16s MiSeq Illumina library concentration and PhiX BioGenomics Sample Prep / Library Generation 19 10-11-2017 11:10 AM
Custom primers and PhiX on MiSeq gleb Illumina/Solexa 14 05-09-2017 09:53 AM
Sequence of PhiX MiSeq primers Oregano Illumina/Solexa 4 09-15-2014 04:46 PM
PhiX control on MiSeq targeted panels. lethalfang Illumina/Solexa 1 01-24-2013 01:39 AM
How much PhiX in MiSeq run? thdybwf Illumina/Solexa 1 11-26-2012 04:01 AM

Reply
 
Thread Tools
Old 11-03-2015, 07:46 AM   #1
fion
Junior Member
 
Location: Munich

Join Date: Nov 2015
Posts: 2
Default PhiX sequences in Miseq Output

I did a run using miseq (v3 chemistry, paired end 2x300, 25% PhiX added). I did a own designed pcr based library preperation.

Now for some of the samples on the flow cell I get very strange sequences that dont fit what I am expecting. After running it through some aligments i noticed that the sequences all are PhiX sequences (Enterobacteria phage S13, Scaffolding protein D, Protein F ect.)
The samples

As I am quite new to the field of sequencing and havent had this befor I was wondering:
1. Does the PhiX library also contain an Index? (And could it be that I by chance gave this index to my samples thus mixing the sequences I want with lots of PhiX sequences)
2. If that is not the case: Why does the Miseq give me the sequences? (I used a sample sheet and specific Indexes.) How can something without an Index be exporter?
3. Those samples did not produce a lot of reads (about 6000-60 000). Possibly my preperation did not add the adapter correctly to those samples and only surrounding noise from the high concentration of PhiX was measured?

Thanks for any advice
fion is offline   Reply With Quote
Old 11-03-2015, 08:08 AM   #2
GenoMax
Senior Member
 
Location: East Coast USA

Join Date: Feb 2008
Posts: 7,079
Default

Strange. You should have no phiX in your data. How was this run processed (on MiSeq, in BaseSpace, off-line with bcl2fastq)?
GenoMax is offline   Reply With Quote
Old 11-03-2015, 12:28 PM   #3
fion
Junior Member
 
Location: Munich

Join Date: Nov 2015
Posts: 2
Default

Demultiplexing was done on the miseq. Adapter sequences were not trimmed after the demultiplexing. No processing other than directly on the miseq was done. (Our sequencer is noot hoked up to network and we donīt work with BaseSpace)
For pairing I used MiTools (assemble first and overlap consensus sequences) in one evaluation phase and Flash in a control evaluation.
fion is offline   Reply With Quote
Old 11-03-2015, 02:10 PM   #4
GenoMax
Senior Member
 
Location: East Coast USA

Join Date: Feb 2008
Posts: 7,079
Default

I can't think of a scenario where phiX will massively contaminate your sample sequences after demultiplexing. What fraction of reads are mapping to phiX and is the match across the entire read/exact? Is it present in all demultiplexed samples?
GenoMax is offline   Reply With Quote
Old 11-04-2015, 08:32 AM   #5
WhatsOEver
Senior Member
 
Location: Germany

Join Date: Apr 2012
Posts: 215
Default

1) no it doesn't contain an index
2) Are the indices correctly specified in your samplesheet.csv?
3) 6k reads? That is indeed almost nothing, you have had probably a problem during library prep or your pooling calculations weren't very good.

Could you also give some info about the run metrics, like cluster density, >Q30 percentages, etc.
WhatsOEver is offline   Reply With Quote
Old 11-04-2015, 08:38 AM   #6
pmiguel
Senior Member
 
Location: Purdue University, West Lafayette, Indiana

Join Date: Aug 2008
Posts: 2,317
Default

This always happens. Some of the phiX reads bleed over into the indexed data sets.

I think what is going on is that since the phiX clusters are "dark" during indexing they end up, a small fraction of the time, being "illuminated" by a nearby cluster. Thus they are identified as a member of that data set.

If the phiX cluster were producing its own signal it wouldn't be subject to this illumination as the signal it was producing would be much brighter than the nearby cluster's stray signals.

I actually preferred the v2 phiX libraries as they were indexed. Of course that meant that you could only use 23 indexes instead of 24 per lane if you wanted phiX. But it seemed worth it to me.

--
Phillip
pmiguel is offline   Reply With Quote
Old 11-20-2015, 02:47 AM   #7
JBKri
Member
 
Location: Heraklion, Greece

Join Date: Jan 2014
Posts: 79
Default

If the PhiX library is constructed like other Illumina libraries, they should have the same adapter sequences at the ends, where the sequencing primers and indexing primers anneal. During index reads the index primers should anneal and read part of the flowcell adapter sequence for the PhiX clusters. Unless PhiX libraries are somehow different.
Jon


Quote:
Originally Posted by pmiguel View Post
This always happens. Some of the phiX reads bleed over into the indexed data sets.

I think what is going on is that since the phiX clusters are "dark" during indexing they end up, a small fraction of the time, being "illuminated" by a nearby cluster. Thus they are identified as a member of that data set.

If the phiX cluster were producing its own signal it wouldn't be subject to this illumination as the signal it was producing would be much brighter than the nearby cluster's stray signals.

I actually preferred the v2 phiX libraries as they were indexed. Of course that meant that you could only use 23 indexes instead of 24 per lane if you wanted phiX. But it seemed worth it to me.

--
Phillip
JBKri is offline   Reply With Quote
Old 11-20-2015, 05:58 AM   #8
pmiguel
Senior Member
 
Location: Purdue University, West Lafayette, Indiana

Join Date: Aug 2008
Posts: 2,317
Default

Quote:
Originally Posted by JBKri View Post
If the PhiX library is constructed like other Illumina libraries, they should have the same adapter sequences at the ends, where the sequencing primers and indexing primers anneal. During index reads the index primers should anneal and read part of the flowcell adapter sequence for the PhiX clusters. Unless PhiX libraries are somehow different.
Jon
Yes, I think phiX libraries are made with adapters that lack index priming sites entirely. Brian Bushnell posted the phiX adapter sequences fairly recently. So you could check if you like.

--
Phillip
pmiguel is offline   Reply With Quote
Old 11-20-2015, 07:07 AM   #9
kmcarr
Senior Member
 
Location: USA, Midwest

Join Date: May 2008
Posts: 1,178
Default

Quote:
Originally Posted by pmiguel View Post
Yes, I think phiX libraries are made with adapters that lack index priming sites entirely. Brian Bushnell posted the phiX adapter sequences fairly recently. So you could check if you like.

--
Phillip
The index 1 primer site is just the complement of the read 2 piriming site so since these libraries have a read 2 priming site they by definition they have an index 1 priming site. It may be possible the PhiX control has a different read 2 primer site sequence than standard TruSeq libraries. An PhiX specific read 2 primer would need to be included in all of Illumina's kits, which is plausible I suppose.

Index read 2 is primed by the flowcell bound oligo to which the PhiX control library must be able to anneal so I don't see how you could avoid getting signal from PhiX clusters during index read 2 in dual indexing runs.
kmcarr is offline   Reply With Quote
Old 11-20-2015, 07:41 AM   #10
pmiguel
Senior Member
 
Location: Purdue University, West Lafayette, Indiana

Join Date: Aug 2008
Posts: 2,317
Default

Quote:
Originally Posted by kmcarr View Post
The index 1 primer site is just the complement of the read 2 piriming site so since these libraries have a read 2 priming site they by definition they have an index 1 priming site. It may be possible the PhiX control has a different read 2 primer site sequence than standard TruSeq libraries. An PhiX specific read 2 primer would need to be included in all of Illumina's kits, which is plausible I suppose.

Index read 2 is primed by the flowcell bound oligo to which the PhiX control library must be able to anneal so I don't see how you could avoid getting signal from PhiX clusters during index read 2 in dual indexing runs.
As far as your index read 1 point goes:
Sure, but Illumina sequence primer mixes all have several primers to allow compatibility with the various insert-proximate segments of the various Illumina adapter types.
My contention is that the phiX library adapters are provided for in the read 2 primer mix, but not in the index read1 primer mix.

The index read 2 point you make is much harder to dispute. I would need to actually look at the sequence of the phiX adapters. But cursory searches of seqanswers are not finding the post where Brian disclosed those.

--
Phillip
pmiguel is offline   Reply With Quote
Old 11-20-2015, 07:44 AM   #11
GenoMax
Senior Member
 
Location: East Coast USA

Join Date: Feb 2008
Posts: 7,079
Default

Quote:
Originally Posted by pmiguel View Post
I would need to actually look at the sequence of the phiX adapters. But cursory searches of seqanswers are not finding the post where Brian disclosed those.

--
Phillip
They are in BBMap "resources" directory

Code:
>PhiX_read1_adapter
AGATCGGAAGAGCGGTTCAGCAGGAATGCCGAGACCGATCTCGTATGCCGTCTTCTGCTTGAAA
>PhiX_read2_adapter
AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTAGATCTCGGTGGTCGCCGTATCATTAAAAAA
GenoMax is offline   Reply With Quote
Old 11-20-2015, 08:46 AM   #12
pmiguel
Senior Member
 
Location: Purdue University, West Lafayette, Indiana

Join Date: Aug 2008
Posts: 2,317
Default

Alright. Thanks GenoMax.

So "read2" adapter for phiX libraries is a standard TruSeq read2 (right) adapter. So kmcarr is correct about phiX libraries generating an index read2 if one is doing a dual index run.

The "read1" adapter matches the "Oligonucleotide Sequences for Paired End DNA" adapter in the Illumina "Adapter Sequences" memo. I think these were the original adapter sequence for the very first Illumina "PE libraries". Prior to the advent of indexed libraries?

In any case, I think I'm right here. I don't think a primer that would anneal to this particular read 1 adapter would be present in an index read 1 primer mix.

So, my contention is that:

Sequence an Illumina phiX library -- index 1, read 1: dark -- no seq.

--
Phillip
pmiguel is offline   Reply With Quote
Old 12-01-2015, 09:48 AM   #13
JBKri
Member
 
Location: Heraklion, Greece

Join Date: Jan 2014
Posts: 79
Default

Quote:
Originally Posted by pmiguel View Post
Alright. Thanks GenoMax.

So "read2" adapter for phiX libraries is a standard TruSeq read2 (right) adapter. So kmcarr is correct about phiX libraries generating an index read2 if one is doing a dual index run.

The "read1" adapter matches the "Oligonucleotide Sequences for Paired End DNA" adapter in the Illumina "Adapter Sequences" memo. I think these were the original adapter sequence for the very first Illumina "PE libraries". Prior to the advent of indexed libraries?

In any case, I think I'm right here. I don't think a primer that would anneal to this particular read 1 adapter would be present in an index read 1 primer mix.

So, my contention is that:

Sequence an Illumina phiX library -- index 1, read 1: dark -- no seq.

--
Phillip
This is getting confusing. If the read2 adapter for phiX libraries is a standard TruSeq sequence, then Index1 should not be dark.

According to Lou et al 2013, supporting information, "the adapters on the R2 end of the phiX
V3 library contain a non-TruSeq R2 sequencing primer." But the R2 end is where index 1 is (i7), so your conclusion is still correct; first index read - dark.

Furthermore, the text more or less proves what you are saying about cluster "cross-talk" (also suggested by Kircher et al 2011):
"We found that phiX clusters that received index assignments were strikingly closer to indexed nonphiX clusters than phiX clusters that did not receive index assignments."
And "...quality scores for alleged index reads from phiX clusters are dramatically lower than for nonphiX clusters."

It's a shame nuggets like these are hidden in supporting information.
JBKri is offline   Reply With Quote
Old 12-01-2015, 02:11 PM   #14
Brian Bushnell
Super Moderator
 
Location: Walnut Creek, CA

Join Date: Jan 2014
Posts: 2,707
Default

Hmm... I should mention that maybe the PhiX adapter sequence names are a little misleading. I determined them empirically using BBMerge on PhiX reads. The reads with an insert size shorter than read length have adapter sequence, and so it tracks those reads to generate a consensus. "PhiX_read1_adapter" is the consensus adapter sequence that appears in read 1 of PhiX, meaning it is the reverse-complement of the adapter that is physically connected to read 2. So maybe it would be better if the names were reversed.

Also, I don't know if the poly-A at the end of the adapter sequence is real or just what the sequencing machine assigns after going off the end of the adapter.
Brian Bushnell is offline   Reply With Quote
Old 12-04-2015, 09:38 AM   #15
thermophile
Senior Member
 
Location: CT

Join Date: Apr 2015
Posts: 243
Default

Quote:
Originally Posted by pmiguel View Post
Alright. Thanks GenoMax.
Sequence an Illumina phiX library -- index 1, read 1: dark -- no seq.
This is correct. I recently did a test run but instead of doing 100% phix, I decided to throw on a few random samples that needed resequencing. What I ran was 80% phiX and 20% dual indexed amplicons (cleared this with my FAS). The run failed at the first index because 80% of the clusters were dark. One of those bizarre things that my FAS didn't realize would happen but he'd also never seen anyone try to do an 80% phiX index run.

6000 phiX out of ~10M reads is pretty much nothing. Include a phiX filtering step in your qc
thermophile is offline   Reply With Quote
Reply

Tags
custom built, index primer, phix, unspecific signal

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 02:40 AM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2020, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO