Here, we just wish to point out that a number of “canonical” lists lurk within our database and you may find them useful.
First of all, we have compilations of studies relating to “cytokine storms.” There’s a lot of debate about exactly what constitutes a cytokine storm. We see a number of “survivors vs. non-survivors” studies where folks die of viral infections without dramatic upregulation of cytokines in the blood. At the anecdotal level, here in Thailand, where hundreds of folks routinely die per year from dengue, I’ve encountered doctors who claim they’ve never seen anyone actually die of a “cytokine storm” per se. So here’s how we define “cytokine storm”: it’s when you die from an acute viral infection in rapid fashion. Thus we toss chronic infections like HBV and HIV from the dataset. We also toss Ebola, as the viral infection may be so potent that death is more likely than survivorship.
The “cytokine storm” lists are assembled from a mere 7 blood-based studies: GSE95233, GSE43777, GSE17924, GSE97287, GSE101702, GSE111368, and GSE111368. The upregulation list takes the database ID 118771101, and the downregulation list 118772101. We found that the upregulation list strongly overlaps with a GO “secretory granule” set, so we also constructed an upregulation list with secretory granule components excluded: 116037101.
We have plenty to say about cytokine storms, but we won’t start blathering at this point.
Next, we have some cancer-related lists. Based on more than 50 studies, we constructed broad lists of transcripts/proteins that are up- or downregulated in cancer tissue vs. adjacent tissue: 118765101 and 118766101. We also have lists relating to transcripts that are up- or downregulated in metastatic vs. primary tissue: 118767101 and 118768101.
You may intuit that cancer is so heterogeneous across tissues and cell-types that such lists would not be particularly informative. You'd be wrong. At the level of transcript and protein abundance, the same entities are altered again and again. It’s not surprising at all for us to come across a new cancer study, enter the upregulated (or downregulated) entities into our database, and find that the single best mimic of this cancer study comes from a canonical cancer list, with P-values that may exceed 10^-100. Again, we have plenty to say about cancer…but not now.
Finally, we have “canonically altered under interferon treatment” lists based on 20 studies (folks love to hit cells with interferons, with largely similar effects every time). We’ve excluded type II interferons (interferon gamma) from the list, as these have very different effects than type I/III interferons: 118769101 and 118770101.
In the future, we’ll update our cancer lists with new studies; we see perhaps 2 new studies per month. We could also build lists for specific cancer types, if the volume of data warrants it. Another task is to double-check every metastasis-related study, as comparisons can be made between primary tissues from which metastasis has and hasn’t arisen, as well as primary cancer tissue vs cancer tissue that has actually lodged in distant locales. We don’t expect that the cytokine lists will be updated soon, as the underlying studies are few and far between. We’ll probably never update the interferon lists, as it’s pretty clear what happens when you blast cells with these antiviral compounds.
