Seqanswers Leaderboard Ad

Collapse

Announcement

Collapse
No announcement yet.
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Comparative Time Series Analysis of RNA-Seq data?

    I need help with comparative time-series analysis of RNA-Seq Data. I have an untreated sample, then 3 treatments at day 3, the same 3 treatments at day 6 and then at day 9. I have tag-counts(generated from HT-Seq counts) for all of these samples across time. Could you please suggest some tools to get Differentially expressed genes. I do not wish to perform pair-wise comparisons.

    Thank you so much

    Ashu

  • #2
    DESeq(2), edgeR, limma/voom, etc. Anything that can handle GLMs will work.

    Comment


    • #3
      Great, than you so much.

      Could you or anybody explain this to me with an example? I am new to this and its tough to understand what functions/methods to use.

      Originally posted by dpryan View Post
      DESeq(2), edgeR, limma/voom, etc. Anything that can handle GLMs will work.

      Comment


      • #4
        I expect there's an example of this in one of the vignettes/user guides for a couple of the packages I listed. In general, you just want to fit your data with a GLM of the form "counts ~ time*treatment", assuming that there can be a meaningful time:treatment interaction in your experiment (otherwise, just exchange a plus sign for the asterisk).

        Comment


        • #5
          The pdf(http://bioconductor.org/packages/rel.../doc/DESeq.pdf) from bioconductor does this ->
          > fit1 = fitNbinomGLMs( cdsFull, count ~ libType + condition )
          > fit0 = fitNbinomGLMs( cdsFull, count ~ libType )
          > pvalsGLM = nbinomGLMTest( fit1, fit0 )
          > padjGLM = p.adjust( pvalsGLM, method="BH" )

          I dont understand what are they trying to do here. Below is what I tried with my data

          >glm_design

          samples libType time treatment

          Samp_1 single-end d0 Untreated
          Samp_2 single-end d3 treatment1
          Samp_3 single-end d6 treatment1
          Samp_4 single-end d9 treatment1
          Samp_5 single-end d3 treatment2
          Samp_6 single-end d6 treatment2
          Samp_7 single-end d9 treatment2
          Samp_8 single-end d3 treatment3
          Samp_9 single-end d6 treatment3
          Samp_10 single-end d9 treatment3

          glm_cds = newCountDataSet(combined_file , glm_design)

          glm_cds = estimateSizeFactors(glm_cds)

          # I dont have replicates
          glm_cds = estimateDispersions(glm_cds, method="blind", sharingMode="fit-only" )

          fit1 = fitNbinomGLMs( glm_cds, count ~ time*treatment)

          fit0 = fitNbinomGLMs( glm_cds, count ~ time)

          pvalsGLM = nbinomGLMTest( fit1, fit0 )

          padjGLM = p.adjust( pvalsGLM, method="BH" )


          IS THIS WHAT I AM SUPPOSED TO DO?

          unique(padjGLM)
          [1] 1 NA

          padj values do not seem right??
          *************************************

          Originally posted by dpryan View Post
          I expect there's an example of this in one of the vignettes/user guides for a couple of the packages I listed. In general, you just want to fit your data with a GLM of the form "counts ~ time*treatment", assuming that there can be a meaningful time:treatment interaction in your experiment (otherwise, just exchange a plus sign for the asterisk).
          Last edited by ashuchawla; 07-23-2013, 10:13 AM.

          Comment


          • #6
            # I dont have replicates
            glm_cds = estimateDispersions(glm_cds, method="blind", sharingMode="fit-only" )

            fit1 = fitNbinomGLMs( glm_cds, count ~ time*treatment)
            fit0 = fitNbinomGLMs( glm_cds, count ~ time)
            Well, you sort of have replicates, just not full replicates. You might try the default dispersion method and see how well that fits. For fit0, try "fit- <- fitNbinomGLMs(glm_cds, count~time+time:treatment)". Otherwise, you getting output for things that can vary by both treatment and a time:treatment interaction, which you indicated not wanting.

            BTW, an adjusted p-value of 1 is normal and to be expected. That all of them are 1 is too bad, though doing as I suggested above my change that. You can also try doing some independent filtering (see the genefilter package and the accompanying paper in PNAS), which might improve things further.

            Comment


            • #7
              > glm_cds = estimateDispersions(glm_cds)
              Error in .local(object, ...) :
              None of your conditions is replicated. Use method='blind' to estimate across conditions, or 'pooled-CR', if you have crossed factors.



              Originally posted by dpryan View Post
              Well, you sort of have replicates, just not full replicates. You might try the default dispersion method and see how well that fits. For fit0, try "fit- <- fitNbinomGLMs(glm_cds, count~time+time:treatment)". Otherwise, you getting output for things that can vary by both treatment and a time:treatment interaction, which you indicated not wanting.

              BTW, an adjusted p-value of 1 is normal and to be expected. That all of them are 1 is too bad, though doing as I suggested above my change that. You can also try doing some independent filtering (see the genefilter package and the accompanying paper in PNAS), which might improve things further.

              Comment


              • #8
                Ah, right, you might try pooled-CR.

                Comment


                • #9
                  Thank you so much for your help.

                  This is what it tells me when I try pooled-CR

                  glm_cds = estimateDispersions(glm_cds, method="pooled-CR", sharingMode="fit-only" , modelFormula=count~time+time:treatment)
                  Error in FUN(newX[, i], ...) :
                  No residual degrees of freedom. Most likely the design is lacking sufficient replication.

                  Originally posted by dpryan View Post
                  Ah, right, you might try pooled-CR.

                  Comment

                  Latest Articles

                  Collapse

                  • seqadmin
                    Best Practices for Single-Cell Sequencing Analysis
                    by seqadmin



                    While isolating and preparing single cells for sequencing was historically the bottleneck, recent technological advancements have shifted the challenge to data analysis. This highlights the rapidly evolving nature of single-cell sequencing. The inherent complexity of single-cell analysis has intensified with the surge in data volume and the incorporation of diverse and more complex datasets. This article explores the challenges in analysis, examines common pitfalls, offers...
                    06-06-2024, 07:15 AM
                  • seqadmin
                    Latest Developments in Precision Medicine
                    by seqadmin



                    Technological advances have led to drastic improvements in the field of precision medicine, enabling more personalized approaches to treatment. This article explores four leading groups that are overcoming many of the challenges of genomic profiling and precision medicine through their innovative platforms and technologies.

                    Somatic Genomics
                    “We have such a tremendous amount of genetic diversity that exists within each of us, and not just between us as individuals,”...
                    05-24-2024, 01:16 PM

                  ad_right_rmr

                  Collapse

                  News

                  Collapse

                  Topics Statistics Last Post
                  Started by seqadmin, Yesterday, 06:54 AM
                  0 responses
                  10 views
                  0 likes
                  Last Post seqadmin  
                  Started by seqadmin, 06-14-2024, 07:24 AM
                  0 responses
                  17 views
                  0 likes
                  Last Post seqadmin  
                  Started by seqadmin, 06-13-2024, 08:58 AM
                  0 responses
                  16 views
                  0 likes
                  Last Post seqadmin  
                  Started by seqadmin, 06-12-2024, 02:20 PM
                  0 responses
                  17 views
                  0 likes
                  Last Post seqadmin  
                  Working...
                  X