Re-run DESeq on miRNA with fold difference threshold

[sr320]

Can you run with no log2fc limits?

[kubu4]

Identified four differentially expressed miRNAs:

https://github.com/RobertsLab/project-clam-oa/blob/3b23f7cfcdf9cef627556090518d60d685d881bd/output/03.00-D-sRNAseq-gene-expression-DESeq2/deseq2.fdr-0.05.lfc-0.table.csv

That's just the DEseq2 output data, which only has the ShortStack cluster IDs. I'll grab the actual miRNA IDs shortly.

[kubu4]

Here are the four clusters pulled from the ShortStack Results.txt file. miRNA IDs are the last column.

NW_026853363.1:213466-213568    Cluster_3720    NW_026853363.1  213466  213568  103     1906    84      1.0     +       AGAUAUGUUUGAUAUAUUUGGU  864     9       11      129     1048    458     251     22      Ycte-miR-190;mle-miR-190-5p
NW_026858897.1:43133-43248      Cluster_15325   NW_026858897.1  43133   43248   116     1294    54      0.98    +       UGUUCUACGAGCUGUAGCUGU   685     4       34      796     431     29      0       21      YNA
NW_026859924.1:190953-191067    Cluster_17619   NW_026859924.1  190953  191067  115     852     54      0.012   -       UGUUCUACGAGCUGUAACUGUC  509     4       70      175     593     10      0       22      YNA
NW_026865271.1:194028-194125    Cluster_29018   NW_026865271.1  194028  194125  98      66793   264     1.0     +       UCACAACCUGCUUGAAUGAGGAC 31912   323     43      462     5039    38213   22713   23      Ymle-miR-67-3p

[sr320]

I looked at code... but in case I missed were there any others limits imposed? eg minumum read count, present in x number of samples etc?

[kubu4]

Adjusted p-vale <= 0.05.

Otherwise, just DEseq2 defaults.

[kubu4]

Oh, and I also just used the miRNAs identified by ShortStack.

If this is run on the full data set (i.e. all sRNA-seq transcripts), I think it identifies ~50 differentially expressed transcripts, using the log₂ fold change = 1 and adjusted p-value <= 0.05. Although I've run that, it is not currently in the code. I can easily add this back in, if desired.

[sr320]

Yep include that but remove the fold change threshold. - then maybe that gets a heat map

used the miRNAs identified by ShortStack.

What is difference between set indentified by ShortStack versus "All"

[kubu4]

What is difference between set indentified by ShortStack versus "All"

• Those identified by ShortStack fit all the parameters (length, secondary structure, etc) to be categorized as miRNA.

• "All" are all of the short RNA-seq data - which includes miRNA, piRNA, etc.

EDITED: Fixed typo.

[kubu4]

Using all sRNAs with adjusted p-value <= 0.05 and log₂ fold change = 0, DEseq2 identified 708 differentially expressed sRNAs (NOTE: This includes miRNAs.)

https://github.com/RobertsLab/project-clam-oa/blob/3eb5caaa2d7906f182450877a4df6bad61e5e14e/code/03.00-sRNAseq-gene-expression-DESeq2.md

Heatmap of top 50 sRNAs:

EDITED: Fixed typo.