Initial commit

Former-commit-id: 9d6e79a6c144ac12b3cf3d4fd98631e21ac8a1e6

almanac/results/ensemble_evaluation_results.csv

@@ -0,0 +1,6 @@
+model,model_dir,mean_squared_error,pearson,r2_score,spearman
+Voting Ensemble - DL+LGBM+XGBoost+RF (expr (DGI) + drugs (ECFP4)),,2212.3120070602904,0.7221017963287318,0.4944363164971569,0.622573717038555
+Voting Ensemble - DL+LGBM (expr (DGI) + drugs (ECFP4)),,1963.2341275676647,0.7460456052946498,0.5513562852147237,0.6425326435648623
+Voting Ensemble - Top 3 DL models+LGBM+XGBoost+RF,,1913.9632476724066,0.7602320413982955,0.5626158035149351,0.665438305838923
+Voting Ensemble - Top 5 DL models (in terms of RMSE) +LGBM+XGBoost+RF,,1855.8396665808134,0.7661363779287811,0.5758983656767203,0.6721637288262223
+Voting Ensemble - Top 10 DL models +LGBM+XGBoost+RF,,1827.9162049557135,0.7670281280864641,0.5822795126714855,0.6717024752248725

almanac/results/ml_model_evaluation_results_cellminercdb.csv

@@ -0,0 +1,8 @@
+model,model_dir,C,alpha,dual,epsilon,gamma,l1_ratio,learning_rate,linearsvr__C,linearsvr__dual,linearsvr__epsilon,linearsvr__loss,linearsvr__max_iter,loss,max_depth,max_features,max_iter,min_child_weight,min_samples_leaf,min_samples_split,min_split_gain,n_estimators,nystroem__gamma,nystroem__n_components,rbfsampler__gamma,rbfsampler__n_components,subsample,subsample_freq,tree_method,mean_squared_error,pearson,r2_score,spearman
+ElasticNet - expr (target genes (all interactions)) + drug (ECFP4),../results/2021-09-24_10-09-18,,0.00029321308659580964,,,,0.8905437519154037,,,,,,,,,,100000.0,,,,,,,,,,,,,3927.778164907341,0.3202027918979946,0.10241322621069293,0.3364889819437518
+XGBoost - expr (target genes (all interactions)) + drug (ECFP4),../results/2021-09-27_12-58-08,,,,,1.0762624813641142,,0.015408845208944859,,,,,,,8.0,,,5.0,,,,615.0,,,,,0.7014936347327818,,hist,2793.07700121598,0.6225666006246391,0.3617182872328263,0.5020524990355083
+Random Forest - expr (target genes (all interactions)) + drug (ECFP4),../results/2021-09-27_23-01-20,,,,,,,,,,,,,,,sqrt,,,2.0,5.0,,863.0,,,,,,,,2703.260986069145,0.6314128239972147,0.3822432924356495,0.5639976777051605
+LinearSVR - expr (target genes (all interactions)) + drug (ECFP4),../results/2021-09-30_14-20-39,7.8705199326554025,,False,0.10342495585470976,,,,,,,,,squared_epsilon_insensitive,,,100000.0,,,,,,,,,,,,,3927.762133811722,0.3202214843345729,0.10241688968118234,0.3364583933602925
+LGBM - expr (target genes (all interactions)) + drug (ECFP4),../results/2021-10-26_14-36-02,,,,,,,0.05107911749137478,,,,,,,8.0,,,3.0,,,0.6878242640017198,972.0,,,,,0.6730297748696231,5.0,,2606.248468026513,0.6457029976315116,0.4044128624651009,0.5383413703012222
+RBFSampler + LinearSVR - expr (target genes (all interactions)) + drug (ECFP4),../results/2021-10-27_13-46-54,,,,,,,,1000.0,False,10.0,squared_epsilon_insensitive,100000.0,,,,,,,,,,,,0.0001,150.0,,,,4121.784407232396,0.2432654056734893,0.05807838095406415,0.2588806898155513
+Nystroem + LinearSVR - expr (target genes (all interactions)) + drug (ECFP4),../results/2021-10-27_14-13-20,,,,,,,,13.558053311806814,False,0.019433401328014608,squared_epsilon_insensitive,100000.0,,,,,,,,,,0.0007899335083362674,150.0,,,,,,4080.079650993491,0.2604770811046381,0.06760886766499853,0.2831788582866129

almanac/results/model_evaluation_results_cellminercdb.csv

@@ -0,0 +1,25 @@
+model,model_dir,cnv_dim,cnv_hlayers_sizes,drug_char_dict,drug_concat_heads,drug_dim,drug_dropout,drug_gat_layers,drug_gcn_layers,drug_hlayers_sizes,drug_kernel_sizes,drug_n_embedding,drug_num_attention_heads,drug_num_filters,drug_residual_connection,drug_seq_length,expr_batchnorm,expr_dim,expr_hlayers_sizes,expr_kernel_size,expr_kernel_size1,expr_kernel_size_rest,expr_kernel_sizes,expr_num_filters,expr_pool_size,hidden_activation,hidden_dropout,l2,learn_rate,mut_dim,mut_hlayers_sizes,predictor_hlayers_sizes,keras_pearson,keras_r2_score,keras_spearman,loss,mean_squared_error,root_mean_squared_error
+"one-hot encoded cell line names + drug (ECFP4, 1024, dense)",../results/2021-04-08_12-02-14,,,,,1024.0,,,,[1024],,,,,,,,60.0,[16],,,,,,,prelu,0.4,0.0,0.0001,,,[2048],0.5542824268341064,0.1728974878787994,0.4909752607345581,2876.86669921875,2876.86669921875,53.63642883300781
+one-hot encoded cell line names + one-hot encoded drugs,../results/2021-04-08_22-27-11,,,,,106.0,,,,[64],,,,,,,,60.0,[16],,,,,,,relu,0.1,0.1,0.001,,,[128],0.0080564869567751,-0.0762210562825203,0.0065891211852431,4375.68017578125,4375.4560546875,66.14723205566406
+Baseline (always predicts the mean of the training set),,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,-5.21983767431955e-05,,,4376.15966796875,66.15254974365234
+"expr (landmark genes (REDO), dense, MinMaxScaler) + drug (ECFP4, Dense)",../results/2021-07-06_12-02-29,,,,,1024.0,,,,"[1024, 512]",,,,,,,,978.0,"[512, 256, 128]",,,,,,,relu,0.1,1E-05,0.0001,,,"[1024, 1024, 1024]",0.7078940868377686,0.44929006695747375,0.6396458745002747,2073.822509765625,2073.653076171875,45.53738021850586
+"expr (NCG genes, dense, MinMaxScaler) + drug (ECFP4, Dense)",../results/2021-07-06_12-21-29,,,,,1024.0,,,,[256],,,,,,,,2362.0,[1024],,,,,,,leakyrelu,0.0,0.001,0.0001,,,"[256, 256, 256]",0.6719173192977905,0.4268694519996643,0.5908969640731812,2092.898681640625,2083.66552734375,45.64718627929688
+"expr (target genes (all interactions), dense, MinMaxScaler) + drug (ECFP4, Dense)",../results/2021-07-06_12-19-03,,,,,1024.0,,,,[512],,,,,,,,976.0,"[1024, 512, 256]",,,,,,,prelu,0.1,0.1,0.0001,,,"[512, 256]",0.699948251247406,0.4495596289634705,0.6239503026008606,2207.822021484375,1974.6983642578125,44.43758010864258
+"expr (targets full (all drug-gene interactions) + NCG genes, dense, MinMaxScaler) + drug (ECFP4, Dense)",../results/2021-07-06_12-38-01,,,,,1024.0,,,,[512],,,,,,,,3037.0,[256],,,,,,,prelu,0.4,0.0001,0.001,,,[512],0.662465512752533,0.4196572899818421,0.5955237150192261,2244.37451171875,2204.98388671875,46.95725631713867
+"expr (COSMIC genes, dense, MinMaxScaler) + drug (ECFP4, Dense)",../results/2021-07-06_12-12-35,,,,,1024.0,,,,"[1024, 512, 256]",,,,,,,,713.0,[64],,,,,,,prelu,0.1,0.001,0.0001,,,[2048],0.6974815726280212,0.4583701491355896,0.6230427622795105,1981.4697265625,1965.61181640625,44.33522033691406
+"expr (targets (all interactions) + landmark genes, dense, MinMaxScaler) + drug (ECFP4, Dense)",../results/2021-07-06_12-37-08,,,,,1024.0,,,,[1024],,,,,,,,1815.0,"[64, 32]",,,,,,,prelu,0.1,0.0,0.01,,,"[256, 256]",0.667938768863678,0.4192943572998047,0.5874897837638855,2098.8603515625,2098.8603515625,45.81332015991211
+"expr (UMAP) + drug (ECFP4, 1024, dense)",../results/2021-07-08_09-48-06,,,,,1024.0,,,,[64],,,,,,,,50.0,"[32, 16]",,,,,,,prelu,0.0,0.001,0.001,,,"[1024, 512]",0.6618035435676575,0.37101367115974426,0.5843226313591003,2325.557373046875,2289.27001953125,47.84631729125977
+"expr (WGCNA, dense) + drug (ECFP4 1024, dense)",../results/2021-07-08_10-02-23,,,,,1024.0,,,,[256],,,,,,,,136.0,"[64, 32]",,,,,,,prelu,0.5,0.0001,0.001,,,"[1024, 512]",0.4909670650959015,0.17902125418186188,0.4124475419521332,3261.425048828125,3260.05615234375,57.096900939941406
+"expr (full, clustering order, 1D Conv) + drug (ECFP4, 1024, dense)",../results/2021-07-10_14-02-58,,,,,1024.0,,,,[256],,,,,,,True,18779,,,,,"[5, 5]","[64, 64]",5,leakyrelu,0.0,0.0,0.001,,,"[1024, 1024]",0.6557958126068115,0.3655383288860321,0.5886580348014832,2083.1611328125,2083.1611328125,45.64165878295898
+"expr (full, chromosome position order, 1D Conv) + drug (ECFP4, 1024, dense)",../results/2021-07-10_14-08-47,,,,,1024.0,,,,[128],,,,,,,True,18779,,,,,"[5, 5]","[32, 32]",10,leakyrelu,0.0,0.001,0.0001,,,"[256, 128, 64]",0.6391177177429199,0.3621706962585449,0.5711956024169922,2133.502685546875,2131.23828125,46.16533660888672
+"expr (full, chromosome position order, 2D Conv) + drug (ECFP4, 1024, dense)",../results/2021-07-12_22-15-05,,,,,1024.0,,,,"[128, 64]",,,,,,,True,"(138, 138, 1)",,"(5, 5)",,,,[32],"(2, 2)",prelu,0.0,0.01,0.001,,,[4096],0.6289666295051575,0.3005034923553467,0.5608550906181335,2479.861083984375,2310.5,48.06766128540039
+"expr (full, clustering order, 2D Conv) + drug (ECFP4, 1024, dense)",../results/2021-07-11_19-59-41,,,,,1024.0,,,,[512],,,,,,,False,"(138, 138, 1)",,"(3, 3)",,,,"[16, 32, 64]","(2, 2)",relu,0.1,0.1,0.0001,,,"[256, 128, 64]",0.5528121590614319,0.19783851504325867,0.4915273785591126,2923.986328125,2798.124267578125,52.89730072021485
+"expr (full, dense, MinMaxScaler) + drug (ECFP4, 1024, dense)",../results/2021-07-11_13-02-09,,,,,1024.0,,,,"[512, 256]",,,,,,,,18779,"[256, 128]",,,,,,,leakyrelu,0.2,0.0,0.0001,,,"[1024, 512, 256]",0.6472998261451721,0.36955910921096796,0.5815195441246033,2099.316650390625,2099.316650390625,45.81830215454102
+"expr (target genes (all interactions), dense, MinMaxScaler) + drug (MolecularTransformerEmbeddings, Dense)",../results/2021-07-13_15-43-50,,,,,512.0,,,,[256],,,,,,,,976,[16],,,,,,,leakyrelu,0.0,0.0,0.0001,,,"[512, 512, 512]",0.6721371412277222,0.4145258069038391,0.5959885716438293,2095.568359375,2095.568359375,45.77737808227539
+"expr (target genes (all interactions), dense, MinMaxScaler) + drug (LayeredFP, Dense)",../results/2021-07-13_20-03-54,,,,,1024.0,,,,[1024],,,,,,,,976,[64],,,,,,,leakyrelu,0.0,0.0001,0.001,,,"[2048, 1024, 512]",0.6776489615440369,0.4284853637218475,0.6013271808624268,2116.98486328125,2050.58154296875,45.2833480834961
+"expr (target genes (all interactions), dense, MinMaxScaler) + drug (GCN)",../results/2021-07-14_22-20-49,,,,,96.0,0.0,,"[128, 128]",,,,,,1.0,,,976,"[512, 256, 128]",,,,,,,prelu,0.3,0.001,0.001,,,"[2048, 2048]",0.6838496923446655,0.4051499664783478,0.6086046695709229,2459.685546875,2201.314697265625,46.91817092895508
+"expr (target genes (all interactions), dense, MinMaxScaler) + drug (TextCNN)",../results/2021-07-14_22-20-12,,,"{'#': 1, '(': 2, ')': 3, '+': 4, '-': 5, '/': 6, '1': 7, '2': 8, '3': 9, '4': 10, '5': 11, '6': 12, '7': 13, '8': 14, '=': 15, 'C': 16, 'F': 17, 'H': 18, 'I': 19, 'N': 20, 'O': 21, 'P': 22, 'S': 23, '[': 24, '\\': 25, ']': 26, '_': 27, 'c': 28, 'Cl': 29, 'Br': 30, 'n': 31, 'o': 32, 's': 33, 't': 34, '.': 35, 'B': 36, '9': 37, 'A': 38}",,214.0,0.5,,,,"[3, 5, 7]",75.0,,"[32, 32, 32, 32, 64, 64, 64, 64, 128, 128, 128, 128]",,214.0,,976,"[256, 128]",,,,,,,prelu,0.2,0.1,0.001,,,[512],0.6140351295471191,0.31435805559158325,0.5489282011985779,2562.740234375,2315.595947265625,48.12063980102539
+"expr (target genes, dense, MinMaxScaler) + mut (gene-level, target genes) + cnv (GISTIC, target genes) + drug (ECFP4, Dense)",../results/2021-07-20_12-00-33,952.0,"[32, 16]",,,1024.0,,,,[512],,,,,,,,976,"[128, 64]",,,,,,,prelu,0.0,0.0001,0.001,636.0,[256],[256],0.6579149961471558,0.3941482305526733,0.5800743699073792,2212.72412109375,2204.220703125,46.94912719726562
+"expr (target genes, dense, MinMaxScaler) + mut (pathway-level, target genes) + cnv (GISTIC, target genes) + drug (ECFP4, Dense)",../results/2021-09-09_15-05-24,952.0,"[256, 128, 64]",,,1024.0,,,,[512],,,,,,,,976,[64],,,,,,,prelu,0.1,0.0,0.0001,1510.0,[256],[1024],0.6853049993515015,0.4380902349948883,0.6131357550621033,2048.585693359375,2048.585693359375,45.26130294799805
+"expr (target genes (all interactions), dense, MinMaxScaler) + one-hot encoded drugs",../results/2021-09-09_22-00-33,,,,,106.0,,,,[8],,,,,,,,976,[128],,,,,,,prelu,0.5,0.001,0.001,,,"[256, 128, 64]",0.011891916394233705,-0.0297666285187006,0.014207258820533752,4326.19873046875,4323.5859375,65.75398254394531
+"expr (target genes (all interactions), dense, MinMaxScaler) + drug (GAT)",../results/2021-09-10_13-40-49,,,,1.0,96.0,0.0,"[8, 8, 8]",,,,,6.0,,0.0,,,976,"[64, 32, 16]",,,,,,,prelu,0.0,0.0001,0.0001,,,"[1024, 1024, 1024]",0.6047568917274475,0.30929267406463623,0.5265606641769409,2587.9658203125,2587.1455078125,50.863990783691406

almanac/results/shap_analysis/multi_input_shap_explanation.pkl.REMOVED.git-id

@@ -0,0 +1 @@
+10ac08568caad56ac309b617b0e27476cca20af6