It's not just negative data.
I think it would be pretty OK to have a thorough and systematic study of a different methodics, preferably finding the good one, but even writing that all we tried wasn't good enough would be perfectly fine. Just done well. Or something.
But this is not the case.
We should be testing transcriptional-based predictive score on snap-frozen samples RNA. They kind of find out, they are not able to secure any reasonable number of such samples for initial testing (like.. more than two, with the actual diagnosis) in due time. So, fallback was to try it on PPFE-isolated RNA. There was a paper where they did this, also, problem was high heterogenity, but when they run hundreds of them and made and split up by median, there still was visible predictive value.
I got 11 samples. Later it was found that two of those samples were from a single person and it was NOT even the specified diagnosis (= useless). And another one was so tiny there was not enough RNA, and next one got some RNA but unusable qPCR values.
So I ended with seven samples I needed to make a median from...
And then.. I wanted to know whether this wacky number even correlate at least with survival or whatever.. well.. they weren't probably even treated with the specified regimen to be able to tell how they responded, in the first place, and patients seem to die randomly even compared with clinical prognosis (which isn't hard to get in sample size = 7).. so I got borderline bad data from unsufficient sample size correlated with nonexistent response.. I would just NOT make any paper from that.
But you can't end a grant with "oh, well, it didn't work and forget about it please.." so instead I'm trying to stitch up something about that you can isolate RNA from PPFE (gee.. you could read dozens of papers on that I suppose, but not one in "local" journal that everyone who can't speak english read or so) and you can even measure some Cts from it but it's not very nice so better not do it. Just from the absence of the "right" way to do it, which I cannot do.
I hate it. I hate it. I hate it.
Of course the original goal was to make this a new local diagnostic tool... well only thing I could write a paper about is how clearly this is impossible here, since all organization and co-operation sucks.. but that probably wouldn't fulfil the grant purpose, right..