Good morning. Please find below a question I have about CuffDiff analysis.

I made a study some months ago with 24 animales, arranged in groups of three.
Now, I have made an additional study which I expected to be more intensive in
terms of differential expression since conditions were stronger, but I have
found no regulated genes (there were hundreds in the previous studies). Just
to check, I repeated a previous study and now I find no regulated gene in a
comparison where I have found more than 200.

Looking in detail, I have found that some genes in one of the control samples
(file genes.readgroup_tracking) have read values of 0 counts (perhaps
suggesting that the bam file has not mapped right). However, many of the other
genes have correct (similar to previous) count numbers. For instanced, in gene

ENSRNOG00000000010

sample
F_sal_veh - replicate 1 | 1
—|—
has 0 counts in this gene. This value is not right, since I have found
positive mapping in this sample as well as in all of the study, inckuding
their biological replicates.

On the contrary, some other genes such as ENSRNOG00000000033 or
ENSRNOG00000000042 (as examples) have identical count records (according to
readgroup_tracking) than the original study. But even in these cases, fold
change values are quite different than in the previous study.

I assume that data in the sample who lacks a number of genes might be
different due to normalization, but I do not understand such big differences
also in terms of FC and q values

I attach the data in an excel file (showing group counts and fold change
values from the original study and in the new one, made with the same bam
files) as well as the cuffdiff log files and command file of the new study. I
would ask you if you might find what the differences may be.

Please let me know if you need further explanation about the studies I have
made.

Regards

Ricardo

Attachments