Using multiple MIDs in titanium sequence runs

[JurgenP]

Hi!

So far I usually pool up to 5 MID-tagged libraries (genomic shotgun) per physically seperated region on a picotiterplate. Thus far this has resulted in up to 4-fold difference in traces per MID (eg: 30 Mb sequenced from 1 MID, 120 MB sequenced from another, where the targeted amount was 75 Mb). These libraries are equal in size distribution.

Have any of you similar experiences and/or suggestions to level the amount of traces per MID, so the amount of sequence data is more equally distributed between MID libraries?

Has someone tried combining 10+ MID tagged libraries in 1 region?

Jurgen

[sulfobus]

Hi
We regularly use up to 10 "in-house MIDs" for 454, and yes, balancing the load is very difficult. We regularly see 4-fold differences, often more. More accurate quantification might help, but I find it hard to believe that that is the only explanation. The emPCR probably introduces some bias as well. I was told by a pro that separate emPCRs for each MID might help.

[kmcarr]

We have also done up to 10 MIDs and too have found wide variations in representation. We are currently quantitating using the Qubit fluorimeter but are considering qPCR.

Originally Posted by sulfobus

I was told by a pro that separate emPCRs for each MID might help.

Not to be a negative Nancy but doesn't that defeat one of the most useful bits of using MIDs? The lab staff has to run only a single, large emPCR.

[sulfobus]

Originally Posted by kmcarr

Not to be a negative Nancy but doesn't that defeat one of the most useful bits of using MIDs? The lab staff has to run only a single, large emPCR.

Indeed, but it saves some money in reagents and plates. The samples we sequence sometimes only contain ~1000 unique sequences, so even a 16th plate would be oversampling. That's why we use ID-tags and pool them.

[JurgenP]

So far I have done some tweaking regarding this "problem". I think factors that influence the askew distribution in MID's are:

1. Concentration of the stock DNA used right before pipetting into the emulsion PCR.
2. Storage of the DNA: both temperature (libraries are single stranded and the DNA strands could form hydrogenbonds when stored to long at 4 degrees or kept too long on the bench at room temperature, both resulting in a reduction of single DNA copies that could end up in a micelle) and type of tubes (maybe an absolute number of DNA could stick to regular Eppendorf tubes - posing problems when storing too long resulting in a lower concentration than expected)

I don't know about the influence of the emPCR: the reagents here are the same for all MID-tagged libraries, and the libraries are similar as well.

Just some thoughts... Any other suggestions that contribute to this fenomenon?

[pmiguel]

Originally Posted by JurgenP

Hi!

So far I usually pool up to 5 MID-tagged libraries (genomic shotgun) per physically seperated region on a picotiterplate. Thus far this has resulted in up to 4-fold difference in traces per MID (eg: 30 Mb sequenced from 1 MID, 120 MB sequenced from another, where the targeted amount was 75 Mb). These libraries are equal in size distribution.

Have any of you similar experiences and/or suggestions to level the amount of traces per MID, so the amount of sequence data is more equally distributed between MID libraries?

Has someone tried combining 10+ MID tagged libraries in 1 region?

Jurgen

Yes, we used 10 MIDs in a single region. We also saw 4-fold differences--anywhere from 6.2 to 23 megabases of sequence for a sample. We did qPCR on each library prior to emPCR.

These were SMART cDNA libraries. Could be the new rapid library technology would produce less varied amounts of sequence.

That said, we were pretty happy with these results. And it was for 10 libraries, so seeing one or two outliers was not unexpected.

--
Phillip

[wchenault]

To add to Jurgen comments about DNA sticking to tubes-- most standard PP tubes bind DNA and other molecules. You should be able to find DNA/RNA lobind tubes in the market

Originally Posted by JurgenP

So far I have done some tweaking regarding this "problem". I think factors that influence the askew distribution in MID's are:

1. Concentration of the stock DNA used right before pipetting into the emulsion PCR.
2. Storage of the DNA: both temperature (libraries are single stranded and the DNA strands could form hydrogenbonds when stored to long at 4 degrees or kept too long on the bench at room temperature, both resulting in a reduction of single DNA copies that could end up in a micelle) and type of tubes (maybe an absolute number of DNA could stick to regular Eppendorf tubes - posing problems when storing too long resulting in a lower concentration than expected)

I don't know about the influence of the emPCR: the reagents here are the same for all MID-tagged libraries, and the libraries are similar as well.

Just some thoughts... Any other suggestions that contribute to this fenomenon?

[ChickSeq]

We have 132 in-house developed Titanium MIDs available for use. I think that the greatest number we have ever pooled in a single region for project data is somewhere around 110. Generally, our projects consist of pools of around 50-75 MID-tagged libraries.

We have definitely noticed that quantitation makes a huge difference. What we'll generally do if we have a large range of concentrations (greater than 2-3 orders of magnitude) is that we'll make two pools--one of "high" concentration libraries and one of "low" concentration libraries. This can help to decrease any bias that we might see in the samples.

[greigite]

I have run 9 genomic libraries prepared with the Nextera kit in one region and saw max 2-fold diffs in yield. These were all pooled prior to emPCR. There was a weak relationship between library concentration as measured with picogreen and total read yield (R-squared 0.1). However, the libraries with the lowest read count were also those for which Bioanalyzer (DNA 7500) and picogreen (PG) quantitation were the most different. For most libs PG and Bioanalyzer agreed well but for the low yield libs there was a 30% difference. In fact, R-squared is 0.75 for the relationship between the ratio (PG conc/Bioanalyzer conc) to read yield (see attached). This suggests to me some inherent property of a few libraries that resulted in inaccurate quantitation with two different methods prior to the emPCR. In the future I'd probably apply both quantitation methods to the libraries, and for those with ratios < 0.8, up the amount of input DNA by ~50% to compensate.

Originally Posted by pmiguel

Yes, we used 10 MIDs in a single region. We also saw 4-fold differences--anywhere from 6.2 to 23 megabases of sequence for a sample. We did qPCR on each library prior to emPCR.

These were SMART cDNA libraries. Could be the new rapid library technology would produce less varied amounts of sequence.

That said, we were pretty happy with these results. And it was for 10 libraries, so seeing one or two outliers was not unexpected.

--
Phillip