You Always Get What You Screen For
Originally published on LinkedIn, June 4, 2026
High throughput screening is often treated as a straightforward exercise. You build an assay around the hits you want to find – binding partners, enzyme activity, phenotypes.
Most of the focus goes toward throughput. How do you screen more candidates, faster?
But assay conditions are not neutral. They shape what biology is allowed to emerge from the screen.
As a great mentor of mine once told me, “You always get what you screen for.”
I often hear teams saying things like:
“We’re not getting the hits we want”
“The screen is broken”
“It’s not working”
The truth is that screens always work, but perhaps not the way you think they do.
I’ve lived this several times.
In grad school, I tried building a DNA-binding screen for transcription factors. Everything bound non-specifically until I stopped looking for proteins that simply bound and started focusing on binding kinetics instead. Throughput dropped, but the biology finally made sense.
At SelectX, an activity assay produced too many nulls. We redesigned the workflow around pooled LC/MS identification and moved activity validation downstream. Throughput increased while true positive identification improved.
Later at Dow AgroSciences, phenotypic screening reinforced the same lesson from another angle. We were screening production strains in microtiter plates. The team had struggled with hits that could not be validated at larger scale. Eventually we realized dissolved oxygen was limiting growth under the assay conditions. Once we adjusted the media composition, the screen produced hits that actually scaled.
The biology was always there. The assay conditions determined whether we could see it.