High-throughput scoring of yeast growth. Well - using 96-well plates, anyway

This is fantastic. Can't wait to get into the lab and try it out! High-resolution Yeast Growth Curves in 96-well Plates Abstract: To compare effects of screen compounds on various yeast strains, YPD yeast cultures were seeded from single colonies and grown overnight to saturation. The resulting cultures were diluted to approximately 1000 cells/µL and grown until log-phase was reached (absorbance of 0.6–0.9 at 600 nm). The log-phase culture cells were collected by centrifugation and washed twice with sterile water to remove the YPD media. The resulting pellet of cells was resuspended in the appropriate liquid media for the experiment. Cultures of 150 µL (1000 cells/µL) were deposited into wells of a 96-well round-bottom polystyrene plate. The plate was lidded and incubated at room temperature with continuous shaking and automated recording of the absorbance (600 nm) every six min. We did a high-throughput screen last summer to evaluate activities of our pilot natural product library against millions of gene-gene interaction pairs in Saccharomyces. The idea was to map any activity of these compounds onto specific gene products (proteins). We have some intriguing leads that, if confirmed, could shed insight on the biology of the producer pathogen. The linked article article confirms my hopes that we can test whether hit activity reproduces in a high-resolution assay.

Amgen scientists find 10% of preclinical reports reproducible

Still a problem: in 2012, Amgen scientists reported that only ~10% of "landmark" cancer precilinical pubs are reproducible. Predicating careers on sexiness of results and novelty, instead of reproducibility, is toxic to science and therefore to everyone who needs results that work.  See also.

"For example, showing data from tumour models in which a drug is inactive, and may not completely fit an original hypothesis, is just as important as showing models in which the hypothesis was confirmed."

That this needs to be broadcast to the field means that the incentive structure is broken. If all your data do not support the story you want to tell, and not for trivial reasons like "my puppy made those aliquots," then these are particularly critical inconsistencies to report. Whether you want to publish in Nature or Cell is and should be irrelevant - the data do not fully support the hypothesis.

In the end, it behooves no one to publish in Nature and temporarily boost your career while dooming months of subsequent postdoc work-hours to rediscovering the irreproducibility of your Nature-approved hypothesis.