maxing out multiplexing in mRNA seq?

ahodgins · 01-05-2011, 05:37 PM

Hi all,

We are designing an experiment in which we are interested in measuring 100+ transcriptomes using Illumina mRNA sequencing. I wonder whether anyone who is currently doing multiplexing for transcriptome analysis has any experience to share about relative depth of coverage at different levels of multiplexing. What would be the maximum number of transcriptomic samples you would be comfortable combining in a single lane with the GAII? (We work with yeast.)

Many thanks,

Andrea

JakobHedegaard · 01-07-2011, 03:31 AM

Hi Andrea,

It depends on how many sequences you will need per sample. If you are happy with 5 M sequences per sample you can combine 7-8 samples per lane (if working on a GAIIx with an expected 35-40 M sequences/lane).

You could also combine more samples (e.g. 20-plex) and then repeat on several lanes (and in different flowcells) to gain enough sequences. This aproach migth be usefull if you would like to account for the minor differences between flowcells/runs.

Cheers,
Jakob

epistatic · 01-08-2011, 05:19 AM

Hi Andrea,

I have done about 32 RNA-Seq libraries for S. cerevisiae. All of the GAIIx data we generate and I have had generated off-site are more in the 20-30M range, ~25M read average per lane. For all of these, I have done 4-plex to balance the bases in the index and have enough coverage for things that pop up like unmapped reads, rRNA contamination, less even coverage from mixing libraries, etc. This will change when we get our HiSeq up and running, I have been told to expect ~120M+ reads per lane and I will be able to do more samples per lane at a lower run cost. I am also moving towards strand-specific and DSN normalized, which may need more reads.

I really like the NuGEN NGS library prep and multiplexing kit. The kit supports 8-plex, but the real benefit is that the index is in the first 4 bp of the read (5' most end). The Illumina indexing needs a second read for the index, either adding cost or reducing your first read. I made an equimolar mix of the four libraries together (BC1 - BC4, BC5 - BC8) and then I sequence in a standard SR-36 run.

epistatic · 01-08-2011, 11:08 AM

Multiplexed yeast mRNA-Seq, Illumina GAIIx 10/2010

Read Counts
Index, Lane 1, Lane 2, BC#
ACCC, 5264469, 28367, BC1
CGTA, 10523020, 77495, BC2
GAGT, 6176660, 60123, BC3
TTAG, 7077502, 199002, BC4
AGGG, 175973, 8185099, BC5
CCAT, 18504, 7832521, BC6
GTCA, 26668, 7003926, BC7
TATC, 7576, 4831595, BC8

% of lane
Index, Lane 1, Lane 2, BC#
ACCC, 16.69%, 0.09%, BC1
CGTA, 33.36%, 0.25%, BC2
GAGT, 19.58%, 0.19%, BC3
TTAG, 22.44%, 0.64%, BC4
AGGG, 0.56%, 26.52%, BC5
CCAT, 0.06%, 25.37%, BC6
GTCA, 0.08%, 22.69%, BC7
TATC, 0.02%, 15.65%, BC8

ahodgins · 01-11-2011, 08:44 AM

Many thanks all!! This was very helpful.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Multiplexing ChIP-seq	Giles	Sample Prep / Library Generation	3	10-26-2011 10:41 AM
BWA parameters for mRNA-seq aligning against mRNA refseq	kwicher	SOLiD	1	09-19-2011 04:45 AM
RNA-seq multiplexing	sppearce	Sample Prep / Library Generation	1	08-05-2011 05:13 AM
mrna seq or directional mrna seq	link1	Sample Prep / Library Generation	0	08-12-2010 06:58 PM
mRNA seq to multiplexing	Chilanga	Illumina/Solexa	1	10-20-2009 04:59 AM

01-05-2011, 05:37 PM	#1
ahodgins Junior Member Location: New Haven, CT Join Date: Jan 2010 Posts: 4	maxing out multiplexing in mRNA seq? Hi all, We are designing an experiment in which we are interested in measuring 100+ transcriptomes using Illumina mRNA sequencing. I wonder whether anyone who is currently doing multiplexing for transcriptome analysis has any experience to share about relative depth of coverage at different levels of multiplexing. What would be the maximum number of transcriptomic samples you would be comfortable combining in a single lane with the GAII? (We work with yeast.) Many thanks, Andrea

01-08-2011, 05:19 AM	#3
epistatic Senior Member Location: Dronning Maud Land Join Date: Mar 2009 Posts: 129	Hi Andrea, I have done about 32 RNA-Seq libraries for S. cerevisiae. All of the GAIIx data we generate and I have had generated off-site are more in the 20-30M range, ~25M read average per lane. For all of these, I have done 4-plex to balance the bases in the index and have enough coverage for things that pop up like unmapped reads, rRNA contamination, less even coverage from mixing libraries, etc. This will change when we get our HiSeq up and running, I have been told to expect ~120M+ reads per lane and I will be able to do more samples per lane at a lower run cost. I am also moving towards strand-specific and DSN normalized, which may need more reads. I really like the NuGEN NGS library prep and multiplexing kit. The kit supports 8-plex, but the real benefit is that the index is in the first 4 bp of the read (5' most end). The Illumina indexing needs a second read for the index, either adding cost or reducing your first read. I made an equimolar mix of the four libraries together (BC1 - BC4, BC5 - BC8) and then I sequence in a standard SR-36 run. Last edited by epistatic; 01-08-2011 at 11:10 AM.

01-08-2011, 11:08 AM	#4
epistatic Senior Member Location: Dronning Maud Land Join Date: Mar 2009 Posts: 129	Results of 4-plex on two GAIIx lanes Multiplexed yeast mRNA-Seq, Illumina GAIIx 10/2010 Read Counts Index, Lane 1, Lane 2, BC# ACCC, 5264469, 28367, BC1 CGTA, 10523020, 77495, BC2 GAGT, 6176660, 60123, BC3 TTAG, 7077502, 199002, BC4 AGGG, 175973, 8185099, BC5 CCAT, 18504, 7832521, BC6 GTCA, 26668, 7003926, BC7 TATC, 7576, 4831595, BC8 % of lane Index, Lane 1, Lane 2, BC# ACCC, 16.69%, 0.09%, BC1 CGTA, 33.36%, 0.25%, BC2 GAGT, 19.58%, 0.19%, BC3 TTAG, 22.44%, 0.64%, BC4 AGGG, 0.56%, 26.52%, BC5 CCAT, 0.06%, 25.37%, BC6 GTCA, 0.08%, 22.69%, BC7 TATC, 0.02%, 15.65%, BC8

01-07-2011, 03:31 AM	#2
JakobHedegaard Member Location: Aarhus, Denmark Join Date: Mar 2008 Posts: 57	Hi Andrea, It depends on how many sequences you will need per sample. If you are happy with 5 M sequences per sample you can combine 7-8 samples per lane (if working on a GAIIx with an expected 35-40 M sequences/lane). You could also combine more samples (e.g. 20-plex) and then repeat on several lanes (and in different flowcells) to gain enough sequences. This aproach migth be usefull if you would like to account for the minor differences between flowcells/runs. Cheers, Jakob

01-11-2011, 08:44 AM	#5
ahodgins Junior Member Location: New Haven, CT Join Date: Jan 2010 Posts: 4	Many thanks all!! This was very helpful.