If you cannot get counts-limma trend it is a viable approach (see the voom article by the limma authors). Please also note that the Bioconductor post you refrence only states you should not use limma-voom on FPKM values. Actually in that exact post Gordon highlights (as option 3) if you only have FPKM you can use limma-trend via a log2 transformation with a pseudocount.
If you have different library sizes you are right they are not comparable (also what the voom article shows) but then you can just do a inter-library normalization of the FPKM counts first (exactly like you are doing for count data).
• 0 views
•
link
written
by
Kristoffer Vitting-Seerup
Hi, You can get read count or No. of mapped reads by using samtools idxstats command, where the 3rd column in the output file is the read count.
The command is :
$ samtools idxstats input.bam > output.txt
(Note: The input bam is the alignment bam file generated by any alignment tool (Hisat2) or assembly tool (Spades) which is coordinate sorted and indexed)
Hope this will help you!!
Hi, You can get read count or No. of mapped reads by using samtools idxstats command, where the 3rd column in the output file is the read count.
The command is :
$ samtools idxstats input.bam > output.txt
(Note: The input bam is the alignment bam file generated by any alignment tool (Hisat2) or assembly tool (Spades) which is coordinate sorted and indexed)
Hope this will help you!!
I think that you may have mis-understood the question. The user does not want total and aligned read counts per sample. They need raw read counts per gene per sample.
Yes you can use limma although It is not as good as if you had counts - so double check you cannot get the counts and try writing the people behind the data to see if they will provide the counts (most people are quite friendly in my experince). If you cannot get the counts you can log2 transform the FPKM values (use a pseudocount of 1) and use the limma trend approach described on page 71 of the limma vignette. For a comparison of limma-trend vs limma-voom take a look at this article.
Hope this helps
Thanks a lot. It really helps☺
Page 71, limma, where do they state you can use rpm/fpkm/tpm for differential expression analysis.
15.4 Differential expression: limma-trend
If the sequencing depth is reasonably consistent across the RNA samples, then the simplest and most robust approach to differential exis to use limma-trend. This approach will usually work well if the ratio of the largest library size to the smallest is not more than about 3-fold. In the limma-trend approach, the counts are converted to logCPM values using edgeR’s cpm function:
The prior count is used here to damp down the variances of logarithms of low counts. The logCPM values can then be used in any standard limma pipeline, using the trend=TRUE argument when running eBayes or treat. For example:
Or, to give more weight to fold-changes in the gene ranking, one might use:
15.5 Differential expression: voom
When the library sizes are quite variable between samples, then the voom approach is theoretically more powerful than limma-trend. In this approach, the voom transformation is applied to the normalized and filtered DGEList object:
The voom transformation uses the experiment design matrix, and produces an EList object. It is also possible to give a matrix of counts directly to voom without TMM normalization, by
If the data are very noisy, one can apply the same between-array normalization methods as would be used for microarrays, for example:
After this, the usual limma pipelines for differential expression can be applied, for example:
Or, to give more weight to fold-changes in the ranking, one could use say:
That is wrong. If you read that closely you see they use cpm. That is a critical difference because CPM is the properly-normalized counts without length correction is fine for limma-trend while length-corrected data such as FPKM completely distort the mean-variance trend which is the whole point of methods such as limma which gain power by sharing information across genes. At no point they recommend FPKM for anything.