The Biological Reality of Cancer
Every cell in your body contains DNA — roughly 3 billion base pairs encoding ~20,000 genes. These genes produce proteins. Proteins do everything — they carry signals, catalyze reactions, control cell division.
Cancer starts when a mutation in DNA causes a protein to malfunction. Specifically, proteins that control cell division — when they mutate, cells divide uncontrollably. That's a tumor.
The mutation isn't random noise. It's a specific, identifiable change. EGFR for example is a receptor protein sitting on the cell surface. When it mutates, it gets permanently stuck in the "ON" position — telling the cell to divide constantly even without a signal to do so.
How Drugs Work Against This
Targeted cancer therapy works by designing a molecule that fits into the mutated protein's binding pocket and blocks it — turns the stuck "ON" switch back off.
Erlotinib does this for EGFR. It sits inside the ATP binding pocket of EGFR and physically blocks the signaling. Cell division slows. Tumor shrinks. Patient responds.
This is the best case scenario in modern oncology — a molecularly targeted drug hitting a specific mutation. It's cleaner than chemotherapy which kills everything dividing, healthy or not.
Why It Falls Apart — The Resistance Problem
Here's where it gets brutal.
A tumor isn't one cell. It's billions of cells, and every time a cell divides there's a chance of another mutation. So within a tumor you don't have one genetic profile — you have thousands of slightly different subpopulations, all competing.
When you give erlotinib, it kills 99.9% of EGFR-mutant cells. But somewhere in that tumor was one cell that had already acquired a second mutation — T790M — which reshaped the ATP binding pocket just enough that erlotinib no longer fits. That cell survives. Divides. In 6-12 months the patient relapses and now the entire tumor is T790M positive.
The drug didn't fail because it was bad science. It failed because the tumor evolved around it.
This happens with virtually every targeted cancer therapy ever developed. It is the central unsolved problem in oncology.
What Oncologists Currently Do About It
When a patient relapses, the standard approach is:
- Biopsy the tumor again
- Send it for genomic sequencing (takes 2-4 weeks)
- Identify what new mutation caused resistance
- Search literature for whether any drug works against that mutation
- If yes — try that drug. If no — try chemotherapy or a clinical trial.
This process is reactive. By the time resistance is detected, the patient has already relapsed. And step 4 often returns nothing — for many resistance mutations, no drug exists yet because the structural characterization hasn't been done.
Where ProtPocket Sits In This Picture
Right now, between step 3 (mutation identified) and step 4 (find a drug), there is a black box. You know the mutation. You don't know what it did to the protein structurally. You don't know if any existing compound could still bind. You don't know if the pocket even exists anymore.
Structural biologists can answer this — but it takes months of crystallography work, specialized equipment, and significant funding. Most hospitals and research groups can't do this for every patient mutation they encounter.
ProtPocket answers that black box question computationally, in minutes, for free, in a browser.
Given a mutation, it tells you:
- What happened to the binding pocket geometrically
- Whether existing drug candidates can still fit
- Whether the pocket collapsed, shrank, or whether a new cryptic pocket appeared elsewhere on the protein
That's the medical need. That's the gap.
Why It's Needed Now Specifically
Tumor genomic sequencing has become cheap and routine. Hospitals are sequencing patient tumors regularly. The bottleneck has shifted — data is no longer the problem. Interpretation is.
Oncologists are drowning in mutation data they can't structurally interpret. A patient comes back with 47 mutations in their relapsed tumor. Which ones matter for druggability? Which ones destroyed the target pocket? Which ones opened new ones?
No accessible tool answers this today. The tools that can (Schrödinger, MOE) cost hundreds of thousands of dollars annually and require PhD-level expertise to operate. Academic structural biology labs that could do this work have years-long queues.
What To Expect In The Future — Medically
Near term (where ProtPocket fits):
Single mutation → pocket impact → existing drug re-evaluation. This is immediately clinically useful for the most common resistance mutations which are already catalogued in COSMIC and ClinVar.
Medium term:
Compound mutation analysis. Real patient tumors don't have one resistance mutation — they have several. T790M + C797S together in EGFR, for example, defeats every approved EGFR inhibitor simultaneously. A tool that takes a full mutation panel from a patient's sequencing report and returns a druggability landscape for each mutation combination — that's the next frontier.
Long term — where this field is heading:
Prospective resistance prediction. Instead of reacting to resistance after it develops, oncologists will want to know before starting therapy — given this patient's tumor mutation profile, which resistance mutations are most likely to emerge, and are there drugs that can preemptively handle both the primary mutation and the likely resistance mutations simultaneously. This is called anticipatory drug design and it's where the field is moving.
AlphaFold + AlphaMissense together make computational approaches to this genuinely feasible for the first time. ProtPocket is building the foundational layer that makes this pipeline accessible.
The one line medical summary:
Cancer evolves around drugs by mutating the proteins they target. We need tools that tell us — in real time, for any mutation, for any researcher — whether that evolution destroyed the drug's target or created a new one. That tool doesn't exist accessibly today. ProtPocket is building it.
The Biological Reality of Cancer
Every cell in your body contains DNA — roughly 3 billion base pairs encoding ~20,000 genes. These genes produce proteins. Proteins do everything — they carry signals, catalyze reactions, control cell division.
Cancer starts when a mutation in DNA causes a protein to malfunction. Specifically, proteins that control cell division — when they mutate, cells divide uncontrollably. That's a tumor.
The mutation isn't random noise. It's a specific, identifiable change. EGFR for example is a receptor protein sitting on the cell surface. When it mutates, it gets permanently stuck in the "ON" position — telling the cell to divide constantly even without a signal to do so.
How Drugs Work Against This
Targeted cancer therapy works by designing a molecule that fits into the mutated protein's binding pocket and blocks it — turns the stuck "ON" switch back off.
Erlotinib does this for EGFR. It sits inside the ATP binding pocket of EGFR and physically blocks the signaling. Cell division slows. Tumor shrinks. Patient responds.
This is the best case scenario in modern oncology — a molecularly targeted drug hitting a specific mutation. It's cleaner than chemotherapy which kills everything dividing, healthy or not.
Why It Falls Apart — The Resistance Problem
Here's where it gets brutal.
A tumor isn't one cell. It's billions of cells, and every time a cell divides there's a chance of another mutation. So within a tumor you don't have one genetic profile — you have thousands of slightly different subpopulations, all competing.
When you give erlotinib, it kills 99.9% of EGFR-mutant cells. But somewhere in that tumor was one cell that had already acquired a second mutation — T790M — which reshaped the ATP binding pocket just enough that erlotinib no longer fits. That cell survives. Divides. In 6-12 months the patient relapses and now the entire tumor is T790M positive.
The drug didn't fail because it was bad science. It failed because the tumor evolved around it.
This happens with virtually every targeted cancer therapy ever developed. It is the central unsolved problem in oncology.
What Oncologists Currently Do About It
When a patient relapses, the standard approach is:
This process is reactive. By the time resistance is detected, the patient has already relapsed. And step 4 often returns nothing — for many resistance mutations, no drug exists yet because the structural characterization hasn't been done.
Where ProtPocket Sits In This Picture
Right now, between step 3 (mutation identified) and step 4 (find a drug), there is a black box. You know the mutation. You don't know what it did to the protein structurally. You don't know if any existing compound could still bind. You don't know if the pocket even exists anymore.
Structural biologists can answer this — but it takes months of crystallography work, specialized equipment, and significant funding. Most hospitals and research groups can't do this for every patient mutation they encounter.
ProtPocket answers that black box question computationally, in minutes, for free, in a browser.
Given a mutation, it tells you:
That's the medical need. That's the gap.
Why It's Needed Now Specifically
Tumor genomic sequencing has become cheap and routine. Hospitals are sequencing patient tumors regularly. The bottleneck has shifted — data is no longer the problem. Interpretation is.
Oncologists are drowning in mutation data they can't structurally interpret. A patient comes back with 47 mutations in their relapsed tumor. Which ones matter for druggability? Which ones destroyed the target pocket? Which ones opened new ones?
No accessible tool answers this today. The tools that can (Schrödinger, MOE) cost hundreds of thousands of dollars annually and require PhD-level expertise to operate. Academic structural biology labs that could do this work have years-long queues.
What To Expect In The Future — Medically
Near term (where ProtPocket fits):
Single mutation → pocket impact → existing drug re-evaluation. This is immediately clinically useful for the most common resistance mutations which are already catalogued in COSMIC and ClinVar.
Medium term:
Compound mutation analysis. Real patient tumors don't have one resistance mutation — they have several. T790M + C797S together in EGFR, for example, defeats every approved EGFR inhibitor simultaneously. A tool that takes a full mutation panel from a patient's sequencing report and returns a druggability landscape for each mutation combination — that's the next frontier.
Long term — where this field is heading:
Prospective resistance prediction. Instead of reacting to resistance after it develops, oncologists will want to know before starting therapy — given this patient's tumor mutation profile, which resistance mutations are most likely to emerge, and are there drugs that can preemptively handle both the primary mutation and the likely resistance mutations simultaneously. This is called anticipatory drug design and it's where the field is moving.
AlphaFold + AlphaMissense together make computational approaches to this genuinely feasible for the first time. ProtPocket is building the foundational layer that makes this pipeline accessible.
The one line medical summary: