This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Differential expressed genes in cuffdiff

Hi,

I got fully confused and frustrated by watching the result

I had 6 samples paired-end reads by tophat2

I used ensemble gtf and pre-index bowtie2

Left reads:
     Input   : 19667353
     Mapped  : 18672982 (94.9% of input)
      of these:  1127159 ( 6.0%) have multiple alignments (939 have >20)
Right reads:
     Input   : 19667353
     Mapped  : 17146234 (87.2% of input)
      of these:  1019710 ( 5.9%) have multiple alignments (848 have >20)
91.1% overall read mapping rate.

Aligned pairs: 16994149
  of these:  1010648 ( 5.9%) have multiple alignments
         25673 ( 0.2%) are discordant alignments
**86.3% concordant pair alignment rate.**

Then I used cuffdiff to get differential expressed genes.

This is my experiment details

In my experiment, before high temperature treatment (48 c), Arabidopsis cotyledons were pre-heated for 2, 8 and 24 hours. In the other hand I have cotyledons keeping in room temperature for 2, 8 and 24 hours, I mean without pre-heating these cotyledons directly face the high temperature. I don have any replication

Something like below

  • pre-heated-2h
  • pre-heated-8h
  • pre-heated-24h
  • unpre-heated-2h
  • unpre-heated-8h
  • unpre-heated-24h

I tried 3 cuffdiff syntaxs,but in ALL, in the gene_exp.diff file the significant column was totally NO, not only this file, the significant column wherever is NO , even when I labeled 6 samples, in the output I have only to columns for samples

cuffdiff -o diff_out -b genome.fa -p 8 \
    -L Pre-2h,unPre-2h,Pre-8h,unPre-8h,Pre-24h,unPre-24h \
    -u merged_asm/merged.gtf \
    -T pre-heated-2h.bam \
    unpre-heated-2h.bam pre-heated-8h.bam \
    unpre-heated-8h.bam pre-heated-24h.bam \
    unpre-heated-24h.bam #syntax1

cuffdiff -o diff_out -b genome.fa -p 8 \
    -L Pre-2h,unPre-2h,Pre-8h,unPre-8h,Pre-24h,unPre-24h \
    -u merged_asm/merged.gtf \
    pre-heated-2h.bam unpre-heated-2h.bam \
    pre-heated-8h.bam unpre-heated-8h.bam \
    pre-heated-24h.bam unpre-heated-24h.bam #syntax2

cuffdiff -o diff_out -b genome.fa -p 8 \
    -L pre,unpre \
    -u merged_asm/merged.gtf \
    pre-heated-2h.bam, pre-heated-8h.bam, \
    pre-heated-24h.bam unpre-heated-2h.bam \
    unpre-heated-8h.bam, unpre-heated-24h.bam #syntax3

cuff = readCufflinks (dbFile = "output_database",
                geneFPKM = "genes.fpkm_tracking",
                geneDiff = "gene_differential_expression_testing",
                isoformFPKM = "isoforms.fpkm_tracking",
                isoformDiff = "transcript_differential_expression_testing",
                TSSFPKM = "tss_groups.fpkm_tracking",
                TSSDiff = "TSS_groups_differential_expression_testing",
                CDSFPKM = "cds.fpkm_tracking",
                CDSExpDiff = "CDS_FPKM_differential_expression_testing",
                 CDSDiff = "CDS_overloading_diffential_expression_testing",
                 promoterFile = "promoters_differential_expression_testing",
                splicingFile = "splicing_differential_expression_testing",
                rebuild = T)
cuff_data <- cuff
csDensity(genes(cuff_data))
csScatter(genes(cuff_data), 'C1','C2')
csVolcano(genes(cuff_data), 'C1','C2')
mygene <- getGene(cuff_data, 'DCL1')
expressionBarplot(mygene)
get_diff_data <- diffData(genes(cuff_data))
> sig_gene_data <- subset(gene_diff_data, (significant == 'yes'))
Error in subset(gene_diff_data, (significant == "yes")) : 
  error in evaluating the argument 'x' in selecting a method for function 'subset': Error: object 'gene_diff_data' not found

How possible that there is no any differential expressed genes in my experiment??? :(

Thank you

rna-seq next-gen
• 3,210 views • link
updated by Ram Ram • written by zizigolu zizigolu

Related note:

Genes are expressed differentially. Deferential = with respect. :)

• 0 views • link
written by Ram Ram

thank you Ram, you all right

• 0 views • link
written by zizigolu zizigolu

We really need more information to tell you what may have went wrong. Some general ideas of the samples and comparisons, what tools and analysis pipeline you used, etc. Something may have gone wrong very early in the process (alignment, trasncript modelling, etc) that is causing your downstream effects.

• 0 views • link
written by DG DG

2 answers

There is "yes" in significance column, I should inspect precisely

• 0 views • link
updated by Ram Ram • written by zizigolu zizigolu

One issue may be that you are currently doing a 6 sample comparison where all pairs are being evaluated against each other. Now that may be what you want to do, but it looks to me like you have a time series experiment with two-conditions. Cuffdiff has the -T flag for doing time series comparisons. Although I have never done this before, so I don't know if you can do time series and differential comparisons between conditions at the same time. You may need to do some sort of two-staged analysis to work this out. At the very least you probably want to do something like Pre-2h vs UnPre-2h, Pre-8h vs UnPre-8h, etc.

You also need to label all samples I believe with the -L flag if you are going to use it. You can't just provide it two labels and 6 samples as you have currently done. Not sure what the downstream result is, if any, but there may be something funny happening there too. Like I said though, not sure if it will impact your downstream results or not.

Have you looked at the actual result files for your comparisons? Those were all output by Cuffdiff, that you are then feeding into CummeRbund. You should check those files to see that you actually have output data and what it looks like as well.

• 0 views • link
updated by Ram Ram • written by DG DG

thank you for your comprehensive explanation

• 0 views • link
written by zizigolu zizigolu

Log in to answer this question.

More posts like this