10 replies to this topic

[ShannonJ]

I am working to sequence negative strand RNA viral genomes de novo (i.e. not the specific-primer based approach that works so very well) and am having trouble with host contamination. In fact I began the project expecting data with ~90% host sequence, but now have ~99.5% host and less than 0.5% target sequence! Since at this rate a full lane of Illumina sequencing is not enough to provide a good draft sequence for a 20Kb genome!

Current protocol (in brief):
1. Virus is grown up in Vero E6 cells and the Trizol cell lysate shipped here
2. Chloroform extraction of the Trizol
3. cDNA synthesis from total RNA
4. Illumina/454 library prep (adaptors added, etc) and sequencing

I've thought about shearing up some monkey (the cell line is Vero E6) DNA, binding it to streptavidin beads. Using those beads to pull out the host DNA and only synthesizing cDNA from the unbound nucleic acid. Unfortunately I’ve never made beads like this (although the idea sounds simple)…any protocol suggestions for bead prep or other avenues to reduce the amount of host sequence in the final output?


Thanks in advance to any responses!

[bob1]

What about doing a virus prep - surely the ultimate way to lower the host cell contamination.  There are plenty of kits out there for it... have a look at the millipore website for some that I can think of straight away.

[ShannonJ]

I'll give it a look, thanks!

bob1, on Sep 29 2009, 05:21 PM, said:

What about doing a virus prep - surely the ultimate way to lower the host cell contamination.  There are plenty of kits out there for it... have a look at the millipore website for some that I can think of straight away.

[ShannonJ]

I don't think the kits are going to work for this particular project. The viruses are BSL3 and 4 and I work in a building with BSL1 and 2 facilities. We have the viruses (negative strand RNA) grown up off-site and then the Trizol cell lysates shipped here. So what I'm trying to figure out is a way to reduce the amount of host contamination in the resulting sequence.

Any other suggestions?

[texsequencer]

What type of adenylated adaptors are you using? I had a similar problem, starting using new custom adaptors and my problem went away. Here are the ones I use: adenylated adaptors

good luck

ShannonJ, on Sep 30 2009, 01:40 PM, said:

I don't think the kits are going to work for this particular project. The viruses are BSL3 and 4 and I work in a building with BSL1 and 2 facilities. We have the viruses (negative strand RNA) grown up off-site and then the Trizol cell lysates shipped here. So what I'm trying to figure out is a way to reduce the amount of host contamination in the resulting sequence.

Any other suggestions?

[ShannonJ]

I haven't bought any yet but was thinking about the Dynabeads M-270 From Invitrogen (cat 653-05). I spoke with a rep from Inv. and while he wasn't very optimistic he did say that it should work but to expect a bunch of optimization to be required.

I'll spend some time looking at the custom adapters you linked to - thanks!

texsequencer, on Oct 30 2009, 03:04 PM, said:

What type of adenylated adaptors are you using? I had a similar problem, starting using new custom adaptors and my problem went away. Here are the ones I use: adenylated adaptors

good luck

Edited by ShannonJ, 02 November 2009 - 07:05 AM.

[Warren]

ShannonJ, on Sep 29 2009, 03:37 PM, said:

I am working to sequence negative strand RNA viral genomes de novo (i.e. not the specific-primer based approach that works so very well) and am having trouble with host contamination. In fact I began the project expecting data with ~90% host sequence, but now have ~99.5% host and less than 0.5% target sequence! Since at this rate a full lane of Illumina sequencing is not enough to provide a good draft sequence for a 20Kb genome!

Current protocol (in brief):
1. Virus is grown up in Vero E6 cells and the Trizol cell lysate shipped here
2. Chloroform extraction of the Trizol
3. cDNA synthesis from total RNA
4. Illumina/454 library prep (adaptors added, etc) and sequencing

I've thought about shearing up some monkey (the cell line is Vero E6) DNA, binding it to streptavidin beads. Using those beads to pull out the host DNA and only synthesizing cDNA from the unbound nucleic acid. Unfortunately I’ve never made beads like this (although the idea sounds simple)…any protocol suggestions for bead prep or other avenues to reduce the amount of host sequence in the final output?


Thanks in advance to any responses!

If you plan on using beads I would definitely not use them the way you are suggesting.  Presumably, your samples are RNA (if not there are certainly ways of getting rid of DNA) -- if you tried to shear up host DNA you have big problems including:  less than 2% of it will be "coding" sequence, of that only 50% will be able to bind RNA (the complementary sequence) and then surely not all the genes are expressed.  So only a tiny percentage of your beads would correspond to any RNA in the sample.  Then you have to take into account that there could be LOTS of certain RNAs yet your only getting a couple sequences per cell of DNA (ie, you can makes lots of mRNA from one gene!)  so you would need tremendous amounts of sequences bound to beads to even make a slight dent in the population of mRNAs in your sample.  If you are stuck on using this, your pretty much have to start with cDNA to even have a chance at working.

However, I think that kind of technology would be better to use to pull out your sequence of interest, rather than remove the huge amount of host nucleic acids.  If you have regions of the virus genome that are well conserved than you could design beads to bind them, wash out all the other crap then elute your viral RNAs.  I don't think those beads need a very large sequence to bind.

Another approach would be to take advantage of the giant size of your RNA of interest.  You could try protocols to specifically precipitate high molecular weight RNA, which you could tweak, or even some kind of column -- since your RNAs are so huge compared to almost anything else in there, you would not have to be very precise to get it to work.  You may even be able to find some spin columns that will not allow your huge RNA through so you can wash out almost everything else.  

warren..

[ShannonJ]

Since pulling out the target isn't really feasible (in the end we want to be able to sequence an unknown virus from a known host), I was actually starting to think along those lines yesterday (size basis of the RNA). Do you think a size exclusion spin column, or something that simple, would work? I ask only because I've never tried to use them on RNA (up until this year I worked in soil microbial community dynamics - largely culture, enzyme and DNA based). Thanks for the tip!
sj

Warren, on Nov 2 2009, 04:11 PM, said:

If you plan on using beads I would definitely not use them the way you are suggesting.  Presumably, your samples are RNA (if not there are certainly ways of getting rid of DNA) -- if you tried to shear up host DNA you have big problems including:  less than 2% of it will be "coding" sequence, of that only 50% will be able to bind RNA (the complementary sequence) and then surely not all the genes are expressed.  So only a tiny percentage of your beads would correspond to any RNA in the sample.  Then you have to take into account that there could be LOTS of certain RNAs yet your only getting a couple sequences per cell of DNA (ie, you can makes lots of mRNA from one gene!)  so you would need tremendous amounts of sequences bound to beads to even make a slight dent in the population of mRNAs in your sample.  If you are stuck on using this, your pretty much have to start with cDNA to even have a chance at working.

However, I think that kind of technology would be better to use to pull out your sequence of interest, rather than remove the huge amount of host nucleic acids.  If you have regions of the virus genome that are well conserved than you could design beads to bind them, wash out all the other crap then elute your viral RNAs.  I don't think those beads need a very large sequence to bind.

Another approach would be to take advantage of the giant size of your RNA of interest.  You could try protocols to specifically precipitate high molecular weight RNA, which you could tweak, or even some kind of column -- since your RNAs are so huge compared to almost anything else in there, you would not have to be very precise to get it to work.  You may even be able to find some spin columns that will not allow your huge RNA through so you can wash out almost everything else.  

warren..

[phage434]

One approach to think about is array based purification.  Roche Nimblegen will make a custom array which can be used to bind the sequences complementary to the array.  This has been used to enrich for rare DNA/RNA.  You'd need to have sequence for the organism you are trying to eliminate.

[ShannonJ]

I had definitely not thought about that - I'll look into it now. Thanks for the tip!

phage434, on Nov 3 2009, 08:00 AM, said:

One approach to think about is array based purification.  Roche Nimblegen will make a custom array which can be used to bind the sequences complementary to the array.  This has been used to enrich for rare DNA/RNA.  You'd need to have sequence for the organism you are trying to eliminate.

[ShannonJ]

Just thought I'd share, I have heard from the Nimblegen folks and they've said the following

"We can only enrich for known sequences.  In addition, I doubt the virus is large enough to justify using our technology over PCR.  She may want to take a look at COLD-PCR.  Not sure it will work, but it might be worth a try."

I don't think that COLD-PCR will be really applicable either but it was definitely a good read!

phage434, on Nov 3 2009, 08:00 AM, said:

One approach to think about is array based purification.  Roche Nimblegen will make a custom array which can be used to bind the sequences complementary to the array.  This has been used to enrich for rare DNA/RNA.  You'd need to have sequence for the organism you are trying to eliminate.