If I use different type of samples in a seq lane - will that introduce a bias?

Sapuizait · 08-13-2020, 02:45 AM

Hi all

I m about to sequence 50 samples (shotgun metagenomics) from animal feces (complex metagenomes) using the novaseq 6000: apx 2.5billion reads in each direction. That way I estimate to have apx 50million reads per sample.
I would like however to add 2 positive controls from pure cultures (probably known E.coli strains that we have in the lab). That way i can check if everything is what it should. By adding 2 extra samples I will then have 52 instead of 50 samples in a single lane which will drop the output to apx 48million reads per sample (I can live with that).
My question is: since often some samples end up having more reads than others, do I introduce a bias in this sequencing run and therefore is it probable that my 2 pos controls will have way more reads than the rest of my 50 samples? if they have less reads, i dont care but if they have much more and i end up having a reduced output for the remaining 50 samples that would be a problem...

thanks
P

GenoMax · 08-13-2020, 03:55 AM

This is all going to depend on how good your libraries are and how well balanced a pool you (or your sequencing provider) can make. There should be no bias in data per se just because of how you pool.

With a large pool of samples I hope you are planning to use dual indexes. If you are particular about getting a balanced pool then running a small iSeq/MiSeq nano run to check the pool balance is well worth doing before final NovaSeq run.

Sapuizait · 08-13-2020, 10:26 AM

the seq provider will take care of the library prep - all I have to do is send the DNA to them, but yes i hope they ll use dual indexes. I honestly hope that they know what they are doing but the truth is I have no choice but to use this company because my collaborator (who will cover 60% of the cost) has an ongoing collaboration with them.

Thus, bottomline is that theoretically, if they use equimolar concentrations of the samples and index it properly, sample type should not be a problem? glad to hear that.

GenoMax · 08-13-2020, 10:38 AM

Rule of thumb is small inserts (and primer dimers) cluster well. So as long as the insert sizes are more or less uniform (and there are no dimers) your results will be as uniform as the pool they can make.

Sapuizait · 08-13-2020, 11:46 AM

ok, got it, thanks! I ll probably contact them before the sequencing to ask about the library prep

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
How many number of samples per lane	smurugesan	Illumina/Solexa	9	08-21-2017 05:02 AM
RNAseq (HighSeq3000) - how many samples per lane for DGE	merrittcr	RNA Sequencing	1	05-02-2017 02:01 PM
multiplexing 16 samples together ran on 4 lanes or multiplexing 4 samples per lane?	zatoichi888	Sample Prep / Library Generation	5	03-30-2014 10:16 AM
Contamination between samples run on the same lane	MLog	Illumina/Solexa	4	12-08-2013 05:42 PM
Can GAIIx introduce a bias in coverage A on T?	vjimenez	Illumina/Solexa	3	05-18-2011 05:11 AM

08-13-2020, 02:45 AM	#1
Sapuizait Junior Member Location: denmark Join Date: Feb 2013 Posts: 3	If I use different type of samples in a seq lane - will that introduce a bias? Hi all I m about to sequence 50 samples (shotgun metagenomics) from animal feces (complex metagenomes) using the novaseq 6000: apx 2.5billion reads in each direction. That way I estimate to have apx 50million reads per sample. I would like however to add 2 positive controls from pure cultures (probably known E.coli strains that we have in the lab). That way i can check if everything is what it should. By adding 2 extra samples I will then have 52 instead of 50 samples in a single lane which will drop the output to apx 48million reads per sample (I can live with that). My question is: since often some samples end up having more reads than others, do I introduce a bias in this sequencing run and therefore is it probable that my 2 pos controls will have way more reads than the rest of my 50 samples? if they have less reads, i dont care but if they have much more and i end up having a reduced output for the remaining 50 samples that would be a problem... thanks P

08-13-2020, 03:55 AM	#2
GenoMax Senior Member Location: East Coast USA Join Date: Feb 2008 Posts: 7,088	This is all going to depend on how good your libraries are and how well balanced a pool you (or your sequencing provider) can make. There should be no bias in data per se just because of how you pool. With a large pool of samples I hope you are planning to use dual indexes. If you are particular about getting a balanced pool then running a small iSeq/MiSeq nano run to check the pool balance is well worth doing before final NovaSeq run. Last edited by GenoMax; 08-13-2020 at 03:57 AM.

08-13-2020, 10:26 AM	#3
Sapuizait Junior Member Location: denmark Join Date: Feb 2013 Posts: 3	the seq provider will take care of the library prep - all I have to do is send the DNA to them, but yes i hope they ll use dual indexes. I honestly hope that they know what they are doing but the truth is I have no choice but to use this company because my collaborator (who will cover 60% of the cost) has an ongoing collaboration with them. Thus, bottomline is that theoretically, if they use equimolar concentrations of the samples and index it properly, sample type should not be a problem? glad to hear that.

08-13-2020, 10:38 AM	#4
GenoMax Senior Member Location: East Coast USA Join Date: Feb 2008 Posts: 7,088	Rule of thumb is small inserts (and primer dimers) cluster well. So as long as the insert sizes are more or less uniform (and there are no dimers) your results will be as uniform as the pool they can make.

08-13-2020, 11:46 AM	#5
Sapuizait Junior Member Location: denmark Join Date: Feb 2013 Posts: 3	ok, got it, thanks! I ll probably contact them before the sequencing to ask about the library prep