update plot

scripts/transformer_paper_plots.py

@@ -462,13 +462,13 @@ def plot_model_load_times_CPU_GPU():
     cpu_load_times = pd.read_csv("../plots/transformer_rnn_runs_model_load_time_final_model_CPU.csv")
     #cpu_load_times["compute_type"] = ["CPU", "CPU", "CPU", "CPU", "CPU"]
 
-    sns.barplot(data=gpu_load_times, x="l_tran", y="tran_load_time", label="", color="green", errorbar="sd", capsize=.2)
-    sns.barplot(data=gpu_load_times, x="l_rnn", y="rnn_load_time", label="", color="red", errorbar="sd", capsize=.2)
-    sns.barplot(data=gpu_load_times, x="l_cnn", y="cnn_load_time", label="", color="blue", errorbar="sd", capsize=.2)
-    sns.barplot(data=gpu_load_times, x="l_dnn", y="dnn_load_time", label="", color="black", errorbar="sd", capsize=.2)
+    #sns.barplot(data=gpu_load_times, x="l_tran", y="tran_load_time", label="", color="green", errorbar="sd", capsize=.2)
+    #sns.barplot(data=gpu_load_times, x="l_rnn", y="rnn_load_time", label="", color="red", errorbar="sd", capsize=.2)
+    #sns.barplot(data=gpu_load_times, x="l_cnn", y="cnn_load_time", label="", color="blue", errorbar="sd", capsize=.2)
+    #sns.barplot(data=gpu_load_times, x="l_dnn", y="dnn_load_time", label="", color="black", errorbar="sd", capsize=.2)
 
-    #sns.barplot(data=cpu_load_times, x="l_tran", y="tran_load_time", label="", linestyle="-", color="green")
-    #sns.barplot(data=cpu_load_times, x="l_rnn", y="rnn_load_time", label="", linestyle="-", color="red")
+    sns.barplot(data=cpu_load_times, x="l_tran", y="tran_load_time", label="", linestyle="-", color="green")
+    sns.barplot(data=cpu_load_times, x="l_rnn", y="rnn_load_time", label="", linestyle="-", color="red")
     #sns.barplot(data=cpu_load_times, x="l_cnn", y="cnn_load_time", label="", linestyle="-", color="blue")
     #sns.barplot(data=cpu_load_times, x="l_dnn", y="dnn_load_time", label="", linestyle="-", color="black")
 
@@ -634,30 +634,35 @@ def plot_usage_time_vs_seq_len():
     plt.savefig("../plots/transformer_rnn_runs_model_pred_time_seq_length.png", dpi=dpi)
 
 
-def make_scatter_beyond_training():
+def make_bar_beyond_training():
 
-    font = {'family': 'serif', 'size': 18}
-    fig_size = (12, 6)
+    #font = {'family': 'serif', 'size': 18}
+    #fig_size = (6, 6)
     #fig = plt.figure(figsize=fig_size)
-    plt.rc('font', **font)
+    #plt.rc('font', **font)
 
-    dpi = 300
-    analysis = "Single-cell"
-    input_tool = "anndata_import"
-    ground_truth = ["scanpy_filter", "anndata_inspect", "anndata_manipulate", "ucsc_cell_browser", "scanpy_inspect", "scanpy_filter_cells"]
+    #dpi = 300
+
+    #ground_truth = ["scanpy_filter", "anndata_inspect", "anndata_manipulate", "ucsc_cell_browser", "scanpy_inspect", "scanpy_filter_cells"]
 
-    pred_transformer_gt = ground_truth
-    pred_rnn_gt = ground_truth
+    #pred_transformer_gt = ground_truth
+    #pred_rnn_gt = ground_truth
 
-    pred_transformer_b_training = ["scanpy_normalise_data", "scanpy_plot", "anndata_ops", "scanpy_remove_confounders", "scanpy_integrate_harmony", "scanpy_normalize", "scpred_get_feature_space", "scanpy_find_variable_genes", "scpred_predict_labels", "scpred_eigen_decompose"]
+    #pred_transformer_b_training = ["scanpy_normalise_data", "scanpy_plot", "anndata_ops", "scanpy_remove_confounders", "scanpy_integrate_harmony", "scanpy_normalize", "scpred_get_feature_space", "scanpy_find_variable_genes", "scpred_predict_labels", "scpred_eigen_decompose"]
 
-    pred_rnn_b_training = ["scanpy_plot", "scanpy_normalise_data", "scmap_scmap_cluster", "scmap_scmap_cell", "scanpy_filter_genes"]
+    #pred_rnn_b_training = ["scanpy_plot", "scanpy_normalise_data", "scmap_scmap_cluster", "scmap_scmap_cell", "scanpy_filter_genes"]
 
     xlabels = ["Transformer", "RNN"]
 
     xtypes = ["Transformer", "RNN"]
+    #analysis = "Proteomics" #"Single-cell"
+    #input_tool = "Proteomics*" #"anndata_import"
 
-    matrix = [len(pred_transformer_b_training), len(pred_rnn_b_training)]
+    matrix = [3, 1]
+    # single cell: (anndata_import) [10, 5]
+    # deep learning: "keras_train_and_eval" [3, 1]
+    # variant calling: "snpeff_sars_cov_2" [5, 3]
+    # proteomics: massspectrometryimagingfiltering, cardinalpreprocessing, cardinalsegmentations [12, 0]
 
     df_recommendations = pd.DataFrame(zip(xlabels, matrix, xtypes), columns=["xlabels", "recommendations", "model_types"])
 
@@ -671,12 +676,33 @@ def make_scatter_beyond_training():
     ax.set_xticks(xlabels)
     plt.xlabel("Model types")
     plt.ylabel("Number of recommended tools")
-    plt.title("Anndata_import: Generalisation")
+    plt.title("Snpeff_sars_cov_2: Generalisation")
+    plt.yticks([0, 2, 4, 6, 8, 10, 12, 14])
+    plt.ylim((0, 15))
     plt.tight_layout()
     plt.savefig("../plots/transformer_rnn_beyond_workflows.pdf", dpi=dpi)
     plt.savefig("../plots/transformer_rnn_beyond_workflows.png", dpi=dpi)
 
 
+def create_test_precision_plot():
+    df_prec = pd.read_csv("../plots/df_tr_rnn_cnn_dnn_runs_te_prec.csv", sep="\t")
+    print(df_prec)
+
+    sns.lineplot(data=df_prec, x="indices", y="tran_prec", label="Transformer", color="green", linestyle="-")
+    sns.lineplot(data=df_prec, x="indices", y="rnn_prec", label="RNN", color="red", linestyle="-")
+    #sns.lineplot(data=df_prec, x="indices", y="cnn_prec", label="CNN", color="blue", linestyle="-")
+    #sns.lineplot(data=df_prec, x="indices", y="dnn_prec", label="DNN", color="black", linestyle="-")
+
+    plt.grid(True)
+    plt.xlabel("Training iterations")
+    plt.ylabel("Precision@k")
+    plt.title("Precision@k of models for test data")
+
+    plt.savefig("../plots/rnn_cnn_dnn_runs_te_prec_defense.pdf", dpi=dpi, bbox_inches='tight')
+    plt.savefig("../plots/rnn_cnn_dnn_runs_te_prec_defense.png", dpi=dpi, bbox_inches='tight')
+    #plt.show()
+
+
 ############ Call methods ###########################
 
 #collect_loss_prec_data(["transformer", "rnn", "cnn", "dnn"])
@@ -685,4 +711,6 @@ def make_scatter_beyond_training():
 #plot_model_load_times_CPU_GPU()
 #plot_usage_time_vs_topk()
 #plot_usage_time_vs_seq_len()
-make_scatter_beyond_training()
+make_bar_beyond_training()
+#create_test_precision_plot()
+#create_ground_truth_beyond_workflows_recommendations_plot()