Fixing Mark03 rule error (#125)

* add wildcard restriction on log_prob analyze_metric rule

* fixed mark03

* change paths to euer

* fix ambiguites

* change initial trees mark02 - were true trees

workflows/analyze.smk

@@ -50,6 +50,10 @@ rule get_log_probs:
     """Extract log probabilities from MCMC samples for ease of plotting / analysis."""
     input:
         mcmc_samples="{DATADIR}/{experiment}/mcmc/MCMC_{mcmc_seed,\d+}-{mutation_data_id}-i{init_tree_id}-{mcmc_config_id}.json",
+    wildcard_constraints:
+        mcmc_config_id = "MC_(?:(?!/).)+",
+        init_tree_id = "(HUN|T)_(?:(?!/).)+"
+
     output:
         result="{DATADIR}/{experiment}/analysis/MCMC_{mcmc_seed,\d+}-{mutation_data_id}-i{init_tree_id}-{mcmc_config_id}/log_prob.json",
     run:

workflows/mark02.smk

@@ -56,7 +56,7 @@ CS_seed = 42 # <-- configure cell simulation seed here
 #####################
 # True Tree Parameters
 tree_types = ["r"] # <-- configure tree type here ["r","s","d"]
-tree_seeds = [42, 34] # <-- configure tree seed here
+tree_seeds = [42]#, 34] # <-- configure tree seed here
 
 #####################
 #####################
@@ -66,8 +66,8 @@ tree_seeds = [42, 34] # <-- configure tree seed here
 # given each error rate, true tree, no of cells and mutations
 initial_points = [ # (mcmc_seed, init_tree_type, init_tree_seed)
     (42, 'r', 45),
-    (12, 'r', 34),
-    (34, 'r', 42),
+    (12, 'r', 20),
+    (34, 'r', 31),
     (79, 'r', 89),
 ]
 

workflows/mark03.smk

@@ -704,4 +704,39 @@ rule combined_logProb_iteration_plot_noHuntress:
         ax.legend(loc="upper right")
 
         # save the histogram
-        fig.savefig(Path(output.combined_logP_iter))
+        fig.savefig(Path(output.combined_logP_iter))
+
+rule get_log_probs_2:
+    """Extract log probabilities from MCMC samples for ease of plotting / analysis.
+        Saves into true tree directory. - not nessary but good for workflow design
+
+        Note this rule is needed only for the MARK03 experiment due to diffuculties with
+        wildcards in the output path.
+        """
+    input:
+        mcmc_samples="{DATADIR}/{experiment}/mcmc/MCMC_{mcmc_seed,\d+}-{mutation_data_id}-i{init_tree_id}-{mcmc_config_id}.json",
+    wildcard_constraints:
+        mcmc_config_id="MC_(?:(?!/).)+",
+        init_tree_id="(HUN|T)_(?:(?!/).)+"
+
+    output:
+        result="{DATADIR}/{experiment}/analysis/MCMC_{mcmc_seed,\d+}-{mutation_data_id}-i{init_tree_id}-{mcmc_config_id}/T_{base_tree_type}_{n_nodes,\d+}_{base_tree_seed,\d+}/log_prob.json",
+    run:
+        # load the data
+        mcmc_samples = yg.serialize.read_mcmc_samples(Path(input.mcmc_samples))
+        mcmc_data = yg.analyze.to_pure_mcmc_data(mcmc_samples)
+
+        # write the result
+        fp = Path(output.result)
+
+        # get log probs
+        log_probs = mcmc_data.log_probabilities
+        # convert log probs to list of float
+        log_probs = [float(i) for i in log_probs]
+        # get iteration
+        iteration = mcmc_data.iterations
+        # convert iterations of list of int
+        iteration = [int(i) for i in iteration]
+
+        # save
+        yg.serialize.save_metric_result(iteration,log_probs,fp)

workflows/mark_viz.smk

@@ -18,4 +18,13 @@ rule mark_viz:
     input:
         f"{DATADIR}/{experiment}/plots/T_r_10_42_relabeled.svg",
         f"{DATADIR}/{experiment}/plots/T_d_10_42_relabeled.svg",
-        f"{DATADIR}/{experiment}/plots/T_s_10_relabeled.svg"
+        f"{DATADIR}/{experiment}/plots/T_s_10_relabeled.svg",
+        f"{DATADIR}/{experiment}/plots/T_r_6_42_relabeled.svg",
+        f"{DATADIR}/{experiment}/plots/T_r_51_42_relabeled.svg",
+        f"{DATADIR}/{experiment}/plots/T_r_6_45_relabeled.svg",
+        f"{DATADIR}/{experiment}/plots/T_r_6_20_relabeled.svg",
+        f"{DATADIR}/{experiment}/plots/T_r_6_34_relabeled.svg",
+        f"{DATADIR}/{experiment}/plots/T_r_6_89_relabeled.svg",
+
+
+

workflows/tree_inference.smk

@@ -18,8 +18,8 @@ from pyggdrasil.tree_inference import (
 ###############################################
 ## Relative path from DATADIR to the repo root
 
-REPODIR = "/cluster/work/bewi/members/gkoehn/repos/PYggdrasil"
-#REPODIR = ".."
+#REPODIR = "/cluster/work/bewi/members/gkoehn/repos/PYggdrasil"
+REPODIR = ".."
 #DATADIR = "/cluster/work/bewi/members/gkoehn/data"
 
 ###############################################

workflows/visualize.smk

@@ -101,6 +101,8 @@ rule plot_tree:
     """Plot a raw tree"""
     input:
         tree="{DATADIR}/{experiment}/trees/{tree_id}.json",
+    wildcard_constraints:
+        tree_id = "^(?!.*_relabeled)(HUN|T)_(?:(?!/).)+$" # allowing both generated and huntress trees
     output:
         plot="{DATADIR}/{experiment}/plots/{tree_id}.svg",
     run: