effect_sizes.txt file empty about leafcutter HOT 7 CLOSED

Interesting: we haven't tried one vs. one (the smallest we've tested is 4 vs. 4), I'll try to check this soon.

from leafcutter.

Ok thank you for looking at this !
Best

from leafcutter.

To run with just one sample both the min_samples_per_intron and min_samples_per_group need to be set to 1. You can do this with e.g.
./leafcutter_ds.R -g 1 -i 1 ...
I would take the p-values when you have no replicates with a (large) grain of salt. Out of curiosity I ran brain vs. muscle comparing N=220 to N=2 (i.e. no replicates as in your setting), and the p-values for both are certainly highly correlated (see attached).

from leafcutter.

I've also added a check that the sample size of the smallest group is not below min_samples_per_intron or min_samples_per_group.

from leafcutter.

Thank you very much. I will give it a try.
Best

from leafcutter.

I've tried the program (on Mac El Capitan) with my 2 samples KO vs WT and checked for the presence of my positive control (floxed exon) found it with p > 0.07, as you said it was expected because of no replicate. I will try with a triplicate. excellent performance. Would be good to have the genomic coordinate on the figure also the gene name sometimes is too close to the border need a small shift to the right. Thank you very much.
Best

from leafcutter.

Thanks for the feedback! I agree those changes would be nice for the plots, I'll let you know if I get that done.

from leafcutter.