AlphaFold2

Deepmind

---

ColabFold

• Google Colab 의 Cloud 서버를 이용해서 Alphafold를 사용.
• deepmid에서 Colab platform을 이용하여 AlphaFold2 script 및 사용법 upload
• Google 서버(GPU)를 사용하여 AlphaFold2 실행 및 결과 확인

Colab AlphaFold2

Process

1. input protein sequence
2. Install dependencies
3. MSA options (custom MSA upload, single sequence, pairing mode)
4. Advanced settings
5. Run Prediction
6. Display 3D structure
7. Plots
8. Package and download results

---

1. input protein sequence

protein sequence (in uniprot)
jobname
templete_mode : none

---

2. Install dependencies

---

3. MSA options (custom MSA upload, single sequence, pairing mode)

mmseqs2_uniref_env (Default)
unpaired_paired (Default)

---

4. Advanced settings

model_type : auto (Default)
num_recycles : auto (Default)

---

5. Run Prediction

5 model output (대략적인 model)

---

6. Display 3D structure

check 3D structure (choose rank_num (1~5))
show_sidechain
color : lDDT, rainbow
blue : Very high
sky : Confident
green : OK
yellow : Low
red : Very low

---

7. Plots

Sequence coverage
Predicted lDDT per position : 100점에 가까우면 정확한 예측에 가깝다.

---

8. Package and download results

---

Summary

protein sequence -> predicted 5 structure