Filtering differentially expressed genes with CummeRbund

[yangruialex]

Hi,

I am new to RNA-seq analysis and I am trying to analysis some data by the tuxedo pipeline.

My samples are control cells and cells with siRNA treatment, each with 4 replicates so 8 samples in total. There are ~40 million reads in each sample. After running these data through cuffdiff, it generates a list of ~9000 genes that are differentially expressed significantly. I went through the list and found that a lot of them have extremely low FPKM value (less than 0.1) and very low fold changes (less than 1.3).

So I have been filtering these with an arbitrary threshold of FPKM>2 and fold change>1.5. I did that in excel and that got me about 900 genes left. My question is that how can I do that in cummeRbund so that I can plot heatmap and other plots with the filtered results?

Thank you very much!

[TiborNagy]

cuff <- readCufflinks()
gene.diff <- diffData(genes(cuff))
filtered <- gene.diff[gene.diff$log2_fold_change > 1.5 & gene.diff$value_1 > 2,]

value_1, value_2 and so on is your samples.

[yangruialex]

cuff <- readCufflinks()
gene.diff <- diffData(genes(cuff))
filtered <- gene.diff[gene.diff$log2_fold_change > 1.5 & gene.diff$value_1 > 2,]

value_1, value_2 and so on is your samples.

That worked! Thank you so much!

[gareth.lim]

Quick question,
If I have 6 conditions (C1-C6), with C1 being my control, do i type in "...gene.diff$ C1 > 2 ,3,4,5,6,]?

[sghignone]

hi!
this thread is the closest I've found to my issue ...

I have 2 conditions (a control and a treatment), each with 3 replicates. How can I do if I want to get those genes showing FPKM > 1, for each replicate, for control OR treatment?

I mean that, for example, a gene must have in control at least fpkm = 1, for all replicates #1, #2 and #3, even if it doesn't happen for the treatment; and vice versa.

This actually works:

gene.diff.filtered <- gene.diff[gene.diff$value_1 > 1 & gene.diff$value_2 > 1,]

but it's not what i want since the fpkm is calculated over the three replica together, and not separately.


Thanks for any hint!
Cheers

stefano

[kwoolf]

I used this to filter my 10,000+ DEGs to a more reasonable number so I could plot heatmaps with the filtered results. Now I can't figure out how to use the filtered results to plot a heatmap. I have tried a few different ways, but none are successful - I cannot use getSig function or get the column of gene_ids from the filtered table into a list to use the getGenes function, so am not sure what other approach to take. I would appreciate if you could share how you did this. Thanks very much!