This lesson is only available in Korean for now.

Dr. Son's Docking and Molecular Dynamics, First Steps

One molecule chosen out of 45,000.
How far should you trust it?

Virtual screening is a funnel: cheap methods throw most things away, expensive methods run only on the few that remain. But whatever the early steps discard never comes back — and the early steps are the least accurate.
These 20 lessons follow ZINC → LigPrep → HTVS → ADMET → Vina → SP → XP → GROMACS → MM-PBSA on one paper's real numbers, working out why a score of −20.66 kcal/mol is not a free energy, why a tannin came first, and why 484 kcal/mol cannot be a binding energy.

🐍 Python🔗 docking🧊 GROMACS ⚖️ MM-PBSA💊 ADMET🚩 false positives🇬🇧 English

Start lesson 1 →

How you'll learn

🔗One stage of the funnel at a time

45,000 → 7,779 → 300 → 50 → 10. At every stage we ask what was thrown away, and on what grounds.

📄Real numbers from a real paper

gscore −20.66 from Table 2, logP 3.43 from the ADMET section, 200 ns / 20 ns from the MD — the values as published.

🚩How to fool yourself, first

Divide the score by size and the winner becomes last. Tannins, PAINS, sign errors, values outside any possible range — you only stop being fooled by working them out yourself.

🛠️Learning by handling it

Every lesson has a laboratory widget — slide a cut-off, raise the flexibility, extend the trajectory, switch MM-PBSA terms on and off.

Full curriculum

45,000개에서 하나로 — 깔때기의 모양

Nath 2017 「Ligand Preparation」~「Protein-Ligand Docking」

ZINC — 살 수 있는 분자들의 목록

「the ligands were downloaded from the ZINC database」

리간드 준비 — 분자 하나가 여럿이 되는 이유

「the ligands were prepared and optimized using LigPrep」

도킹이 푸는 두 문제 — 자세와 점수

어디에 어떻게 놓을 것인가, 그리고 그게 얼마나 좋은가

kcal/mol 인데 자유에너지가 아니다

Table 2 「Glide gscore」 −20.66 kcal/mol 을 어떻게 읽나

무엇을 움직이게 둘 것인가 — 강체·유연 리간드·유연 단백질

「In XP docking, flexibility of the interacting regions were considered」

HTVS → SP → XP — 깔때기를 세 번 좁힌다

「a cut off docking score value of −7 kcal/mol」

Vina 의 10.9 와 Glide 의 −20.66 — 견줄 수 없는 두 눈금

「Only ligands with an affinity value of 10.9 and above were considered」

Fig. 7 읽는 법 — 2D 상호작용 그림에 없는 것

Fig. 7 「Hydrogen bond interactions within the Protein and ligand 15 complex」

약물다움 — 리핀스키 다섯 규칙과 그 예외

Fig. 6 「Parameters used to screen ligands based on drug likeliness」

ADMET — 붙는 것과 약이 되는 것은 다르다

FAF-Drugs3 (Mobyle server) · 7,779개가 남은 자리

탄닌이 1등을 했다 — 위양성의 교과서적 사례

「Beta-1,2,3,4,6-Penta-O-Galloyl-D-Glucopyranose … a naturally occurring tannin」

분자동역학이 도킹에 없는 것을 준다 — 시간

GROMACS · 단백질 200 ns · 복합체 20 ns

상자 짓기 — 물·이온·힘장

「SPC/E water model and AMBER99SB-ILDN protein force-field … 3 Cl atoms were added」

에너지 최소화와 평형화 — 왜 먼저 눌러 주나

「energetically minimized for 0.1 ns and for a maximum force ≥ 10.0 kJ/mol」

RMSD 그림 읽는 법 — 이 논문은 평평해지지 않았다

Fig. 8a · 「the deviation … continues to increase for the full simulation length」

RMSF 와 경첩 잔기 — 어디가 흔들리나

Fig. 8b·c · 최대 요동 209·210·212–215·256·344·345 · 경첩 195·221·274·310·322·332·361

수소결합 수 · SASA · 2차 구조 — 나머지 그림들

「higher numbers of hydrogen bonds between the protein residues than with non-protein residues」

【실습】 MM-PBSA — 200~484 kcal/mol 이 왜 이상한가

「the top hundred ligands based on binding energy with energy in the range of 200–484 kcal/mol」

정리 — 인실리코 결과를 실험으로 넘기는 문턱

무엇이 있으면 이 논문이 강해졌을까