Before declaring that we got biology wrong, examine the gap between what we could measure and what we claimed we could explain.
https://media.damagebdd.com/5301d16331abeef53f05df291f9e4…
Reading DNA, predicting disease, establishing causation and developing a safe treatment are separate scientific achievements. Every transition requires evidence.
The computational gap is enormous.
With one million genetic variants, an exhaustive search involves approximately 500 billion pairs and 167 quadrillion triples. Add interactions with environment, tissue, developmental history and time, and the challenge extends beyond processing more sequences.
It becomes a question of whether our observations can distinguish competing explanations at all.
Stochastic methods are legitimate tools. They can uncover deterministic relationships. But a search method cannot guarantee that the available data contain enough information to identify a causal mechanism.
A correlation can support prediction while leaving the mechanism unresolved. Understanding a mechanism can still leave us without a safe intervention.
This is the gap I am exposing: the distance between acquiring biological data and earning the conclusions drawn from it.
The Human Genome Project achieved its central sequencing objective. The promises surrounding genomic medicine deserve scrutiny wherever they compressed that distance into an assumed technological inevitability.
Computational complexity, limited samples, confounding and causal uncertainty belong inside the promise, budget and timeline from the beginning.
They should also remain visible when we assess the results.
Biology never owed us a shortcut from sequence to cure.
#Genomics #ComputationalBiology #CausalInference #SystemsBiology #ScientificAccountability
https://media.damagebdd.com/5301d16331abeef53f05df291f9e4…
Reading DNA, predicting disease, establishing causation and developing a safe treatment are separate scientific achievements. Every transition requires evidence.
The computational gap is enormous.
With one million genetic variants, an exhaustive search involves approximately 500 billion pairs and 167 quadrillion triples. Add interactions with environment, tissue, developmental history and time, and the challenge extends beyond processing more sequences.
It becomes a question of whether our observations can distinguish competing explanations at all.
Stochastic methods are legitimate tools. They can uncover deterministic relationships. But a search method cannot guarantee that the available data contain enough information to identify a causal mechanism.
A correlation can support prediction while leaving the mechanism unresolved. Understanding a mechanism can still leave us without a safe intervention.
This is the gap I am exposing: the distance between acquiring biological data and earning the conclusions drawn from it.
The Human Genome Project achieved its central sequencing objective. The promises surrounding genomic medicine deserve scrutiny wherever they compressed that distance into an assumed technological inevitability.
Computational complexity, limited samples, confounding and causal uncertainty belong inside the promise, budget and timeline from the beginning.
They should also remain visible when we assess the results.
Biology never owed us a shortcut from sequence to cure.
#Genomics #ComputationalBiology #CausalInference #SystemsBiology #ScientificAccountability
