Disseration
2024-09-12 10:20:46 0 举报
毕业论文流程图
作者其他创作
大纲/内容
20 poses for active ligands
Seven Feature Dataset:span style=\"font-size: inherit;\
GWOVina docking
For all native actives poses:Poses with 2.0Å < RMSD < 4.5Å excluded Poses with 25μm < IC50 < 250μm excludedPoses with RMSD>4.5Å labelled 0Poses with Kd/Ki ≥ 25μm or IC50 ≥ 250 μm labelled 0Poses with RMSD<2.0 Å or Kd/Ki or IC50≤25μm labelled 1Max 3 poses labelled 0 and 5 poses labelled 1
Total:8161 complexs
Total:5102 complexsTrian: 4131 complexesValidation:460 comlplexesTest:511 complexs
1. Combined dataset and removed duplicates2. Removed the problematic and incompatible complexes
For all decoy selected poses:1.BINANA 1.3(531 features);ECIFs(1540 features);Kier flexibility(1 feature)2.Features with 0 variance removed3.Feature pairwise correlation filtering 4.RFECV using random forest(nRot and Kier flexbility not run included in RFECV - added back in afterwards)
1. PDBbind(Refined Set) 2. Binding MOAD(Non-redundant Set) 3. Iridium (Highly Trustworthy Set)
For all decoy:font color=\"#323232\
5 poses for decoy ligand
1. Complex Filtering
2. Decoy generation
6. XGBoost Model Traing
RMSD of native crystal ligand poses calculated
3. Docking
4. Labeling
1.No gird filter: 5102 complexes &51020 decoys2. 1000 grid filter: 5083 complexes & 43797 decoy3. 300 grid filter: 5033 complexes & 26167 decoys4.TocoDecoy and DeepCoy:5102 complexes & 101327 decoys
Label according to self-defined standard data
5. Feature Computation
66688 labeled as 0 : 50373 decoy poses & 16315 active poses 8592 labeled as 1:8592 active posesTotal: 75280 poses
Data Split according to font color=\"#212121\
收藏
0 条评论
下一页
为你推荐
查看更多
抱歉,暂无相关内容