ModDotPlot 0.9.8__tar.gz → 0.9.9__tar.gz

{moddotplot-0.9.8/src/ModDotPlot.egg-info → moddotplot-0.9.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ModDotPlot
-Version: 0.9.8
+Version: 0.9.9
 Author: Michael Schatz, Adam Phillippy
 Author-email: Alex Sweeten <alex.sweeten@nih.gov>
 Maintainer-email: Alex Sweeten <alex.sweeten@nih.gov>
@@ -140,6 +140,9 @@ positional arguments:
 options:
   -h, --help            show this help message and exit
   ```
+
+Note that running `moddotplot -h` might take a while at first! This is because the Python interpreter is compiling source code into the __pycache__ directory. Subsequent runs will use the pre-compiled code and load much faster!
+
 --- 
 
 ## Usage
@@ -207,6 +210,10 @@ Each partition takes into account a fraction of its neighboring partitions k-mer
 
 Modimizer sketch size. Must be lower than window size `w`. A lower sketch size means less k-mers to compare (and faster runtime), at the expense of lower accuracy. Recommended to be kept >= 1000.
 
+`--forward <bool>`
+
+Use forward k-mers only, instead of the default of canonical k-mers. Warning: this will give strand specific output.
+
 `-r / --resolution <int>`
 
 Dotplot resolution. This corresponds to the number of windows each input sequence is partitioned into. Default is 1000. Overrides the `--window` parameter.

{moddotplot-0.9.8 → moddotplot-0.9.9}/README.md

@@ -88,6 +88,9 @@ positional arguments:
 options:
   -h, --help            show this help message and exit
   ```
+
+Note that running `moddotplot -h` might take a while at first! This is because the Python interpreter is compiling source code into the __pycache__ directory. Subsequent runs will use the pre-compiled code and load much faster!
+
 --- 
 
 ## Usage
@@ -155,6 +158,10 @@ Each partition takes into account a fraction of its neighboring partitions k-mer
 
 Modimizer sketch size. Must be lower than window size `w`. A lower sketch size means less k-mers to compare (and faster runtime), at the expense of lower accuracy. Recommended to be kept >= 1000.
 
+`--forward <bool>`
+
+Use forward k-mers only, instead of the default of canonical k-mers. Warning: this will give strand specific output.
+
 `-r / --resolution <int>`
 
 Dotplot resolution. This corresponds to the number of windows each input sequence is partitioned into. Default is 1000. Overrides the `--window` parameter.

{moddotplot-0.9.8 → moddotplot-0.9.9}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "ModDotPlot"
-version = "0.9.8"
+version = "0.9.9"
 requires-python = ">= 3.7"
 dependencies = [
   "pysam",

{moddotplot-0.9.8 → moddotplot-0.9.9/src/ModDotPlot.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ModDotPlot
-Version: 0.9.8
+Version: 0.9.9
 Author: Michael Schatz, Adam Phillippy
 Author-email: Alex Sweeten <alex.sweeten@nih.gov>
 Maintainer-email: Alex Sweeten <alex.sweeten@nih.gov>
@@ -140,6 +140,9 @@ positional arguments:
 options:
   -h, --help            show this help message and exit
   ```
+
+Note that running `moddotplot -h` might take a while at first! This is because the Python interpreter is compiling source code into the __pycache__ directory. Subsequent runs will use the pre-compiled code and load much faster!
+
 --- 
 
 ## Usage
@@ -207,6 +210,10 @@ Each partition takes into account a fraction of its neighboring partitions k-mer
 
 Modimizer sketch size. Must be lower than window size `w`. A lower sketch size means less k-mers to compare (and faster runtime), at the expense of lower accuracy. Recommended to be kept >= 1000.
 
+`--forward <bool>`
+
+Use forward k-mers only, instead of the default of canonical k-mers. Warning: this will give strand specific output.
+
 `-r / --resolution <int>`
 
 Dotplot resolution. This corresponds to the number of windows each input sequence is partitioned into. Default is 1000. Overrides the `--window` parameter.

{moddotplot-0.9.8 → moddotplot-0.9.9}/src/moddotplot/const.py

@@ -1,4 +1,4 @@
-VERSION = "0.9.8"
+VERSION = "0.9.9"
 COLS = [
     "#query_name",
     "query_start",

{moddotplot-0.9.8 → moddotplot-0.9.9}/src/moddotplot/moddotplot.py

@@ -316,6 +316,18 @@ def get_parser():
         type=str,
     )
 
+    static_parser.add_argument(
+        "--forward",
+        action="store_true",
+        help="Enforce forward only k-mers instead of canonical k-mers. Warning: only use if you want strand-specific output!",
+    )
+
+    static_parser.add_argument(
+        "--plot-direction",
+        action="store_true",
+        help="Create a plot containing the direction of each k-mer array (relative to the first array). Arrays with inversions will be highlighted in blue (forward) and pink (reverse).",
+    )
+
     static_parser.add_argument(
         "--colors",
         default=None,
@@ -635,7 +647,10 @@ def main():
     # -----------LOAD SEQUENCES INTO MEMORY-----------
     kmer_list = []
     for i in fasta_list:
-        kmer_list.append(readKmersFromFile(i, args.kmer, False))
+        if args.forward:
+            kmer_list.append(readKmersFromFile(i, args.kmer, False, True))
+        else:
+            kmer_list.append(readKmersFromFile(i, args.kmer, False, False))
     k_list = [item for sublist in kmer_list for item in sublist]
     # Throw error if compare only selected with one sequence.
     if len(k_list) < 2 and args.compare_only:

{moddotplot-0.9.8 → moddotplot-0.9.9}/src/moddotplot/parse_fasta.py

@@ -37,7 +37,9 @@ def extractRegion(seq_name):
     return None
 
 
-def generateKmersFromFasta(seq: Sequence[str], k: int, quiet: bool) -> Iterable[int]:
+def generateKmersFromFasta(
+    seq: Sequence[str], k: int, quiet: bool, fw_only: bool
+) -> Iterable[int]:
     n = len(seq)
     if not quiet:
         progress_thresholds = round(n / 77)
@@ -60,11 +62,12 @@ def generateKmersFromFasta(seq: Sequence[str], k: int, quiet: bool) -> Iterable[
         # Remove case sensitivity
         kmer = seq[i : i + k].upper()
         fh = mmh3.hash(kmer)
-
-        # Calculate reverse complement hash directly without the need for translation
-        rc = mmh3.hash(kmer[::-1].translate(tab_b))
-
-        yield fh if fh < rc else rc
+        if fw_only:
+            yield fh
+        else:
+            # Calculate reverse complement hash directly without the need for translation
+            rc = mmh3.hash(kmer[::-1].translate(tab_b))
+            yield fh if fh < rc else rc
 
 
 def isValidFasta(file_path):
@@ -151,7 +154,9 @@ def printProgressBar(
         print()
 
 
-def readKmersFromFile(filename: str, ksize: int, quiet: bool) -> List[List[int]]:
+def readKmersFromFile(
+    filename: str, ksize: int, quiet: bool, fw_only: bool
+) -> List[List[int]]:
     """
     Given a filename and an integer k, returns a list of all k-mers found in the sequences in the file.
     """
@@ -161,7 +166,9 @@ def readKmersFromFile(filename: str, ksize: int, quiet: bool) -> List[List[int]]
     for seq_id in seq.references:
         print(f"Retrieving k-mers from {seq_id}.... \n")
         kmers_for_seq = []
-        for kmer_hash in generateKmersFromFasta(seq.fetch(seq_id), ksize, quiet):
+        for kmer_hash in generateKmersFromFasta(
+            seq.fetch(seq_id), ksize, quiet, fw_only
+        ):
             kmers_for_seq.append(kmer_hash)
         all_kmers.append(kmers_for_seq)
         print(f"\n{seq_id} k-mers retrieved! \n")