We enjoy using the tools on our site! In the course of this use, we realize that some alterations might deepen the user’s ability to extract information about genes and gene sets of interest. Therefore, we’d like to point out a couple recent tweaks to the site.
First, you’ll now note a new “Emphasize Internal
Significance” option at the bottom of the black, left frame of most of our
tools. What does that mean? Well, when we find that a set contains
significantly altered genes (p<.05) after adjustment, we generally divide
log(fold-change) by these p-values, and sort the data accordingly. If the data
set does not contain significantly altered genes after adjustment, we simply
sort by fold-change. By selecting “Emphasize Internal Significance”, you’ll be
discarding all data sets wherein no genes were significantly altered.*
There are a couple downsides to tossing such “insignificant”
data. First, look here for a discussion of how significance may lurk within
insignificance. Also, in our early days of data collection, we did not
discriminate between significance and insignificance…much of that data was
simply sorted according to fold-change. Therefore, clicking the “Emphasize
Internal Significance” option may discard truly significant data that was not
labeled as such. Over time, we do rebuild the old data sets in our database,
but this is a slow process.
On the other hand, the feature should allow one to better
home in on gene sets of interest. You could use it in conjunction with the “RestrictIDs” feature to really whittle down the data that will be output. You may wish
to ask questions like, “In which studies was my gene significantly downregulated
and found in the top 25 genes within a list?”…with the “Relevant Studies” tool
and proper selection of filters, you can do that!
The second alteration concerns the “Co-expression” tool.
Instead of a single gene name, you can now enter two! In doing so, the
underlying algorithm scans for lists in which both genes (not just one) were
found and outputs a list of genes that tend to also be altered alongside your
two genes. The tweak also applies to “anti-coexpression”; you’ll receive a list
of genes that tend to be altered in the opposite direction to those two genes.
In many cases, two genes are not in the habit of appearing in tandem, so it’s
quite possible to use this new tool and not receive any output. In other words,
it pays to know that your two genes are indeed often co-expressed before you
enter the two genes into the tool.
*Note our wording here. It’s possible that the list of genes
in question contains both significantly and insignificantly altered genes. This
happens when some genes in the list are altered below the adjusted .05
threshold, and others are not.
