Functionmotifs

.github/workflows/action.yml

@@ -52,7 +52,7 @@ jobs:
       - shell: bash -l {0}
         run: mamba install -y -c conda-forge snakemake==7.0 tabulate==0.8.10
       - shell: bash -l {0}
-        run: pip install -e git+https://github.com/PNNL-CompBio/Snekmer@kmer-association#egg=snekmer
+        run: pip install -e git+https://github.com/PNNL-CompBio/Snekmer@functionmotifs#egg=snekmer
 
       #test clustering step
       - name: Snekmer Cluster
@@ -105,3 +105,11 @@ jobs:
           source activate snekmer
           snekmer apply --configfile .test/config_learnapp.yaml -d .test --cores 1
           rm -rf .test/output
+
+      # run Snekmer Motif using previously generated model files
+      - name: Snekmer Motif
+        run: |
+          export PATH="/usr/share/miniconda/bin:$PATH"
+          source activate snekmer
+          snekmer motif --configfile .test/config.yaml -d .test --cores 1
+          rm -rf .test/output

.test/config.yaml

@@ -48,4 +48,6 @@ score_dir: "output/example-model/"
 learnapp:
   save_apply_associations: False
 
-
+# motif params
+motif:
+  n: 200

requirements.txt

@@ -11,4 +11,4 @@ scikit-learn
 tabulate == 0.8.10
 umap-learn
 hdbscan
-pyarrow
+pyarrow

resources/config.yaml

@@ -39,6 +39,10 @@ model:
   random_state: None
 
 # search params
-model_dir: "output/example-model/"
-basis_dir: "output/example-model/"
-score_dir: "output/example-model/"
+model_dir: "output/model/"
+basis_dir: "output/kmerize/"
+score_dir: "output/score/"
+
+# motif params
+motif:
+  n: 2000 

resources/tutorial/demo_example/run_demo.sh

@@ -18,3 +18,6 @@ mv output/scoring/*scorer output/example-model/
 
 # run snekmer search on examples using provided config.yaml
 snekmer search --configfile=../../config.yaml
+
+# run snekmer motif on examples using provided config.yaml
+snekmer motif --configfile=../../config.yaml

snekmer/__init__.py

@@ -8,6 +8,7 @@
 from . import vectorize
 from . import report
 from . import _version
+from . import motif
 
 # from . import walk
 

snekmer/_version.py

@@ -1 +1 @@
-__version__ = "1.2.0"
+__version__ = "1.3.0"

snekmer/cli.py

@@ -257,6 +257,11 @@ def get_argument_parser():
         description="Search sequences against pre-existing models via Snekmer.",
         parents=[parser["smk"]],
     )
+    parser["motif"] = parser["subparsers"].add_parser(
+        "motif",
+        description="Find structurally and functionally relevant motifs via Snekmer.",
+        parents=[parser["smk"]]
+    )
     parser["learn"] = parser["subparsers"].add_parser(
         "learn",
         description="Learn kmer-annotation associations via Snekmer",
@@ -386,6 +391,30 @@ def main():
             verbose=args.verbose,
             quiet=args.quiet,
         )
+
+    elif args.mode == "motif":
+        snakemake(
+            resource_filename("snekmer", os.path.join("rules", "motif.smk")),
+            configfiles=configfile,
+            config=config,
+            cluster_config=args.clust,
+            cluster=cluster,
+            keepgoing=args.keepgoing,
+            force_incomplete=True,
+            forcerun=args.forcerun,
+            cores=args.cores,
+            nodes=args.jobs,
+            workdir=args.directory,
+            dryrun=args.dryrun,
+            unlock=args.unlock,
+            list_code_changes=args.list_code_changes,
+            list_params_changes=args.list_params_changes,
+            until=args.until,
+            touch=args.touch,
+            latency_wait=args.latency,
+            verbose=args.verbose,
+            quiet=args.quiet,
+        )
 
     elif args.mode == "learn":
         snakemake(

snekmer/motif.py

@@ -0,0 +1,109 @@
+"""motif: Identification of structurally and functionally relevant motifs with Snekmer.
+Created on Fri Apr 21 15:25:54 2023
+
+author: @tnitka
+"""
+# ---------------------------------------------------------
+# Imports
+# ---------------------------------------------------------
+# import pickle
+# from datetime import datetime
+
+# import snekmer as skm
+import pandas as pd
+import numpy as np
+# import snekmer.motif
+# from typing import Any, Dict, List, Optional
+# from ._version import __version__
+# from .vectorize import KmerBasis
+# from .score import KmerScorer
+# from .model import SnekmerModel, SnekmerModelCV
+#from numpy.typing import NDArray
+# from sklearn.base import BaseEstimator, ClassifierMixin
+# from sklearn.tree import DecisionTreeClassifier
+# from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier
+# from sklearn.linear_model import LogisticRegression  # LogisticRegressionCV
+# from sklearn.model_selection import GridSearchCV, cross_validate
+# from sklearn.pipeline import make_pipeline, Pipeline
+# from sklearn.svm import SVC
+
+# object to permute training data and retrain
+class SnekmerMotif:
+    """Permute training data and retrain to find highly distinguishing kmers.
+
+    Parameters
+    ----------
+    n : int
+    Number of permutations to test.
+    scores : NDArray
+    """   
+    def __init__(self):
+        self.generator = np.random.default_rng()
+        # self.scorer = skm.score.KmerScorer()
+
+    def permute(self, X: pd.DataFrame, label, label_col="family"):
+        """
+
+        Parameters
+        ----------
+        X : Dataframe containing matrix of shape (n_kmers, n_features)
+            Labeled training data.
+        label : str
+            Primary family label.
+        label_col : str
+            Column with family labels.
+
+        Returns
+        -------
+        Dataframe
+            Training data with permuted labels, for retraining and rescoring.
+
+        """
+        # save primary family label
+        self.primary_label = label
+        self.labels = X[label_col].values
+
+        self.generator.shuffle(self.labels)
+        # self.permuted_labels = self.generator.permutation(self.labels)
+        # self.permuted_data = X
+        X[label_col] = self.labels
+
+        return X
+
+    def p_values(self, X, y: np.ndarray, n: int):
+        """
+
+        Parameters
+        ----------
+        X: Dataframe containing matrix of shape (n_kmers, n_iterations)
+            kmer scores from each permutation tested
+        y: list or array-like of shape (n_kmers, 1)
+            kmer scores from real training data
+        n: int
+            number of permutations tested
+
+        Returns
+        -------
+        Dataframe
+            matrix containing kmer sequences, scores on real data, number of scores
+            on permuted data that exceed that on real data, n_iterations, and
+            proportion of scores on permuted data that exceed that on real data.
+
+        """
+        # self.output = pd.DataFrame(columns=('kmer', 'real score', 'false positives', 'n', 'p'))
+        self.output_matrix = np.empty((1, 5))
+        for i in range(0, len(y)-1):
+            self.seq = X['kmer'].iloc[i]
+            self.real_score = y[i]
+            self.false_score = X.iloc[i, 1:(n+1)].ge(self.real_score).sum()
+            self.p = self.false_score/n
+            self.vec = np.array([[self.seq, self.real_score, self.false_score, n, self.p]])
+            self.output_matrix = np.append(self.output_matrix, self.vec, axis=0)
+
+
+        else:
+            self.output_matrix = np.delete(self.output_matrix, 0, 0)
+
+        self.output = pd.DataFrame(self.output_matrix, columns=('kmer', 'real score', 'false positives', 'n', 'p'))
+
+        return self.output