Misleading extraction of siRNA descriptions as self-activations or inhibitions

[johnbachman]

Hi there, this is an issue that we've known about for a while but I just came across a great example so I thought I would share it. Also with the work going on to extract information about experimental methods (siRNA, knockout, chemical inhibition, etc.) it might now be possible to actually solve, either at the reader level or the assembly level.

The issue is that when an author says "We found that FOO expression was efficiently reduced by FOO siRNA" what they're basically saying is that their attempt to experimentally reduce FOO protein expression by introducing FOO siRNA worked as expected. However, REACH often extracts these (very commonplace) statements as "FOO activates/increases FOO" or (if it misses the polarity indicator, e.g., "siRNA"), "FOO inhibits/decreases FOO."

Of course, according to basic molecular biology, FOO RNA does technically "increase" FOO protein, but given that we don't generally distinguish between RNA and protein abundances this distinction is lost, and these statements can't be distinguished from e.g., feedback regulation at the protein level. These types of statements often pick up a large evidence count because they are quite common.

I've attached a screenshot of a statement with several sentences all showing the same issue. Incidentally the reason some sentences appear multiple times is because "stathmin" and "Op18" are synonyms; the authors write "stathmin/Op18" in the paper, this is expanded by REACH into "stathmin and Op18" and this leads to a combinatorial product of the different relations (another issue to discuss).

Here are the four individual sentences giving rise to "STMN1 inhibits STMN1":

Stmn1 expression level was markedly enhanced by PCA-Stmn1 (P < 0.01), and remarkably decreased by Stmn1 siRNA (P < 0.01; XREF_SUPPLEMENTARY).

To examine the role of stathmin and Op18 in HGF induced WAVE2 transport and lamellipodia formation, stathmin and Op18 expression was depleted by stathmin and Op18 siRNA.

Contrary to these, ultracentrifugation of the cell lysates (D) revealed that suppression of stathmin and Op18 expression by stathmin and Op18 siRNA (50nM) caused a significant increase in the ratio of tubulin polymers to monomers in serum starved cells compared to that in control serum starved cells [<.01] (E).

We demonstrated that lentiviral mediated STMN1 short hairpin RNA (shRNA) specifically and efficiently downregulated STMN1 expression in Eca109 and TE-1 cells.

[MihaiSurdeanu]

Thanks @johnbachman!

A lot of the biology is over my head, so, bringing this back to a language I understand, what should be the correct output in this case?

You are right that we are probably now in a position where we can address this issue, as this type of information is now captured and attached to the corresponding regulation/activation events.