Announcement

Collapse

Welcome to the New Seqanswers!

Welcome to the new Seqanswers! We'd love your feedback, please post any you have to this topic: New Seqanswers Feedback.
See more
See less

Newbie - experimental design question

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Newbie - experimental design question

    We are new to NGS and have a project that we need to set up. In general, our collaborators are looking for sequencing an RNA plant virus from small (25-30 mg) leaf samples that have been stored in a herbarium. The goal is to do some evolutionary calculations based on the sequences that are pulled out of the samples. We can isolate RNA from the sample but only get a 2.3-2.5 RIN with a yield of around 5 micrograms of total RNA from the extraction (this will include plant rRNA & mRNA, chloroplast & mitochondrial RNAs, as well as viral RNAs). Based on preliminary work, the RNA is intact enough to perform RT-PCR for fragments up to 1000 bp. I don't know the titer level of the virus in these plants.

    Any advice on whether we can proceed with RNAseq for Illumina sequencing? What sequence length would be most appropriate? Would GAIIx or HiSeq2000 make a difference? Also, we are looking to pool samples together - any advice? Any other input would also be appreciated!

    Thanks,

    Nikki

  • #2
    One of the first things you do with RNA while making libraries is fragmentation, so you certainly can sequence it. While you can sequence RNA with RINs that low, but it will affect the type and quality of reads you get. What RNA are you interested in sequencing? If you want mRNA, then you will have problems removing other types of RNA. For example, if you do polyA selection with fragmented mRNA, your sequences will be heavily 3' biased. That might be ok for doing just gene level quantification, but that's about it.

    Is there anyway you can get better quality RNA?

    Comment


    • #3
      Thanks pbluescript for the response. I wish there were a way to improve the RNA quality, but the plant samples have undergone a high heat treatment and have been sitting in a herberium for >100 years. The project is looking to obtain full-length virus sequence for a small RNA viral genome.

      What do you think of skipping the rRNA removal step in the Illumina process and simply fragment then ligate the adapters?

      Comment


      • #4
        100 years you say? And I thought I had difficult samples...

        Probably a stupid question, but are you sure that this is RNA? Do you DNase treat your samples before doing RT-PCR?

        I don't think you should fragment your RNA at all. You can certainly try taking the RNA through library preparation, especially since you have a rather high starting amount. If you sequence the samples as they are, you will probably get only a small fraction of the reads coming from the virus. If you have the resources though, it seems like a problem you can buy your way out of, especially since you are just trying to assemble a small RNA viral genome. Just keep sequencing until you get what you want. Definitely do the sequencing on a HiSeq to get as many reads as possible.

        When you said you wanted to pool samples, do you mean multiplex them so you can sequence more than one on a lane? With the samples you have, that might be problematic. What you should do is do a test run so you can determine the percentage of reads you have that originate from the virus and how well your assembly works. Then you can make an estimate of the total number of reads you'll need and how many samples you can multiplex in a lane.

        Comment


        • #5
          Check to see what your Bioanalyser traces should look like for your plant.
          Some plants generally give strange traces (especially some brasicca's) that result in low RINs but the RNA is actually fine. Good RINs are based on having 18s and 28s peaks and everthing else baseline - deviation from baseline lowers the RIN but this could be normal.

          http://corelabs.cgrb.oregonstate.edu...na_quality.pdf

          Comment


          • #6
            Originally posted by pbluescript View Post
            100 years you say? And I thought I had difficult samples...
            I know . . . daunting

            Originally posted by pbluescript View Post
            Probably a stupid question, but are you sure that this is RNA? Do you DNase treat your samples before doing RT-PCR?
            Yes - DNase treated. The virus is a RNA virus with no DNA stage, but to be certain there is no genomic contaminant, it is DNase treated.


            Originally posted by pbluescript View Post
            When you said you wanted to pool samples, do you mean multiplex them so you can sequence more than one on a lane? With the samples you have, that might be problematic. What you should do is do a test run so you can determine the percentage of reads you have that originate from the virus and how well your assembly works. Then you can make an estimate of the total number of reads you'll need and how many samples you can multiplex in a lane.
            I had an idea to do a test run on more modern samples to look at this, but wasn't sure if it was going to be worth it.

            Thanks for the input.

            Comment


            • #7
              Originally posted by TonyBrooks View Post
              Check to see what your Bioanalyser traces should look like for your plant.
              Some plants generally give strange traces (especially some brasicca's) that result in low RINs but the RNA is actually fine. Good RINs are based on having 18s and 28s peaks and everthing else baseline - deviation from baseline lowers the RIN but this could be normal.

              http://corelabs.cgrb.oregonstate.edu...na_quality.pdf
              Thanks for the info, I am not sold on RIN as a quality for plant or insect samples, but my collaborator insists on having at least an RIN of 8 before the NGS samples are run. Your information helps back up my thoughts.

              Comment


              • #8
                One option you could try is DSN normalization to reduce the amount of plant rRNA and other very highly represented sequences. It can be tricky to get it to work well, but a search on seqanswers should help you with the protocol. Since you are not looking to do differential expression analysis, this might be a good choice for your project.

                Comment


                • #9
                  Originally posted by pbluescript View Post
                  One option you could try is DSN normalization to reduce the amount of plant rRNA and other very highly represented sequences. It can be tricky to get it to work well, but a search on seqanswers should help you with the protocol. Since you are not looking to do differential expression analysis, this might be a good choice for your project.
                  The RNA virus we are working with has some very highly structured elements involving large stem-loops. I worry that this protocol will remove critical segments of our viral genome of interest. Any thoughts?

                  Thanks again for all of your input!

                  Comment


                  • #10
                    Originally posted by enkia View Post
                    The RNA virus we are working with has some very highly structured elements involving large stem-loops. I worry that this protocol will remove critical segments of our viral genome of interest. Any thoughts?

                    Thanks again for all of your input!
                    I don't think that will be a concern. First, your RNA will be fragmented due to its age, and the stem loops probably won't be an issue at all. Second, the DSN protocol is only done after the libraries are made, so RNA secondary structure is no longer an issue.

                    Comment


                    • #11
                      We do this kind of work (sequencing ssRNA plant virus genomes) routinely using 454. We multiplex to make it more cost effective and have had entire viral genomes with very few sequences (<10,000 even). If you're after a complete genome that is assembled easily de novo, avoid any kind of cDNA amplification (e.g. SMART) unless you want to get large coverage level of small regions of the genome. It can be done cheaply with 454, Illumina for this seems like overkill unless you're wanting to massively resequence a viral genome which you already have.

                      Comment

                      Working...
                      X