update cstag.to_html to represent repeat substitutions

akikuno · Sep 14, 2022 · 4bad010 · 4bad010
1 parent 79e5b6a
commit 4bad010
Show file tree

Hide file tree

Showing 6 changed files with 313 additions and 13 deletions.
diff --git a/setup.py b/setup.py
@@ -5,7 +5,7 @@
 
 setuptools.setup(
     name="cstag",
-    version="0.2.3",
+    version="0.3.0",
     author="Akihiro Kuno",
     author_email="[email protected]",
     description="Python module to manipulate the minimap2's CS tag",

diff --git a/src/cstag/to_html.py b/src/cstag/to_html.py
@@ -1,20 +1,18 @@
 import re
 
 
-def to_html(CSTAG: str, OUTPUT_FILE_NAME: str, DESCRIPTION: str = "") -> None:
+def to_html(CSTAG: str, DESCRIPTION: str = "") -> None:
     """Output HTML file showing a sequence with mutations colored
     Args:
         CSTAG (str): cs tag in the **long** format
-        OUTPUT_FILE_NAME (str): output file name
         DESCRIPTION (str): (optional) header information in the output file
     Return:
         HTML file (*OUTPUT_FILE_NAME.html*)
     Example:
         >>> import cstag
         >>> CSTAG = "cs:Z:=AC+GGG=T-ACGT*at~gt10cg=GNNN"
-        >>> OUTPUT = "Report"
         >>> DESCRIPTION = "Example"
-        >>> cstag.to_html(CSTAG, OUTPUT, DESCRIPTION)
+        >>> cstag_html = cstag.to_html(CSTAG, DESCRIPTION)
         https://user-images.githubusercontent.com/15861316/158910398-67f480d2-8742-412a-b528-40e545c46513.png
     """
     if not re.search(r"[ACGT]", CSTAG):
@@ -109,12 +107,18 @@ def to_html(CSTAG: str, OUTPUT_FILE_NAME: str, DESCRIPTION: str = "") -> None:
     list_cs = [i + j for i, j in zip(list_cs[0::2], list_cs[1::2])]
 
     html_cs = []
-    for cs in list_cs:
+    idx = 0
+    while idx < len(list_cs):
+        cs = list_cs[idx]
         if cs[0] == "=":
             cs = re.sub(r"(N+)", r"<span class='Unknown'>\1</span>", cs)
             html_cs.append(cs[1:])
         elif cs[0] == "*":
-            html_cs.append(f"<span class='Sub'>{cs[2].upper()}</span>")
+            html_cs.append(f"<span class='Sub'>{cs[2].upper()}")
+            while idx < len(list_cs) - 1 and list_cs[idx+1].startswith("*"):
+                html_cs.append(f"{list_cs[idx+1][2].upper()}")
+                idx += 1
+            html_cs.append("</span>")
         elif cs[0] == "+":
             html_cs.append(f"<span class='Ins'>{cs[1:].upper()}</span>")
         elif cs[0] == "-":
@@ -124,6 +128,7 @@ def to_html(CSTAG: str, OUTPUT_FILE_NAME: str, DESCRIPTION: str = "") -> None:
             splice = "-" * int(cs[3:-2])
             right = cs[-2:].upper()
             html_cs.append(f"<span class='Splice'>{left + splice + right}</span>")
+        idx += 1
 
     html_cs = "".join(html_cs)
     html_cs = f"<p class='p_seq'>{html_cs}</p>"
@@ -137,6 +142,4 @@ def to_html(CSTAG: str, OUTPUT_FILE_NAME: str, DESCRIPTION: str = "") -> None:
             html_footer,
         ]
     )
-
-    with open(OUTPUT_FILE_NAME + ".html", "w", newline="\n") as f:
-        f.write(report)
+    return report
diff --git a/tests/data/to_html/report_substitution.html b/tests/data/to_html/report_substitution.html
@@ -0,0 +1,87 @@
+<!DOCTYPE html>
+<html>
+
+<head>
+    <style>
+        h1 {
+            font-family: Consolas, 'Courier New', monospace;
+            color: #333;
+            padding: 0.1em 0;
+            border-top: solid 3px #333;
+            border-bottom: solid 3px #333;
+        }
+
+        .p_seq {
+            font-family: Consolas, 'Courier New', monospace;
+            color: #585858;
+            word-wrap: break-word;
+            letter-spacing: 0.15em;
+        }
+
+        .p_legend {
+            font-family: Consolas, 'Courier New', monospace;
+            color: #585858;
+            word-wrap: break-word;
+        }
+
+        .Ins {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #ee827c;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+        .Del {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #a0d8ef;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+        .Sub {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #98d98e;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+        .Splice {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #f8e58c;
+            font-weight: bold;
+        }
+
+        .Unknown {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #c0c6c9;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+    </style>
+</head>
+
+<body>
+
+    <h1>Example</h1>
+
+    <p class="p_legend">
+        Labels:
+        <span class="Ins">Insertion</span>
+        <span class="Del">Deletion</span>
+        <span class="Sub">Substitution</span>
+        <span class="Splice">Splicing</span>
+        <span class="Unknown">Unknown</span>
+    </p>
+    <hr>
+
+    <p class='p_seq'>A<span class='Sub'>TG</span>A</p>
+
+</body>
+
+</html>
diff --git a/tests/data/to_html/report_substitution_end.html b/tests/data/to_html/report_substitution_end.html
@@ -0,0 +1,87 @@
+<!DOCTYPE html>
+<html>
+
+<head>
+    <style>
+        h1 {
+            font-family: Consolas, 'Courier New', monospace;
+            color: #333;
+            padding: 0.1em 0;
+            border-top: solid 3px #333;
+            border-bottom: solid 3px #333;
+        }
+
+        .p_seq {
+            font-family: Consolas, 'Courier New', monospace;
+            color: #585858;
+            word-wrap: break-word;
+            letter-spacing: 0.15em;
+        }
+
+        .p_legend {
+            font-family: Consolas, 'Courier New', monospace;
+            color: #585858;
+            word-wrap: break-word;
+        }
+
+        .Ins {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #ee827c;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+        .Del {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #a0d8ef;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+        .Sub {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #98d98e;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+        .Splice {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #f8e58c;
+            font-weight: bold;
+        }
+
+        .Unknown {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #c0c6c9;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+    </style>
+</head>
+
+<body>
+
+    <h1>Example</h1>
+
+    <p class="p_legend">
+        Labels:
+        <span class="Ins">Insertion</span>
+        <span class="Del">Deletion</span>
+        <span class="Sub">Substitution</span>
+        <span class="Splice">Splicing</span>
+        <span class="Unknown">Unknown</span>
+    </p>
+    <hr>
+
+    <p class='p_seq'>A<span class='Sub'>TG</span></p>
+
+</body>
+
+</html>
diff --git a/tests/data/to_html/report_substitution_start.html b/tests/data/to_html/report_substitution_start.html
@@ -0,0 +1,87 @@
+<!DOCTYPE html>
+<html>
+
+<head>
+    <style>
+        h1 {
+            font-family: Consolas, 'Courier New', monospace;
+            color: #333;
+            padding: 0.1em 0;
+            border-top: solid 3px #333;
+            border-bottom: solid 3px #333;
+        }
+
+        .p_seq {
+            font-family: Consolas, 'Courier New', monospace;
+            color: #585858;
+            word-wrap: break-word;
+            letter-spacing: 0.15em;
+        }
+
+        .p_legend {
+            font-family: Consolas, 'Courier New', monospace;
+            color: #585858;
+            word-wrap: break-word;
+        }
+
+        .Ins {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #ee827c;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+        .Del {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #a0d8ef;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+        .Sub {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #98d98e;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+        .Splice {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #f8e58c;
+            font-weight: bold;
+        }
+
+        .Unknown {
+            color: #333;
+            border: 0.1em solid;
+            background-color: #c0c6c9;
+            font-weight: bold;
+            # border-radius: 5px;
+        }
+
+    </style>
+</head>
+
+<body>
+
+    <h1>Example</h1>
+
+    <p class="p_legend">
+        Labels:
+        <span class="Ins">Insertion</span>
+        <span class="Del">Deletion</span>
+        <span class="Sub">Substitution</span>
+        <span class="Splice">Splicing</span>
+        <span class="Unknown">Unknown</span>
+    </p>
+    <hr>
+
+    <p class='p_seq'><span class='Sub'>TG</span>A</p>
+
+</body>
+
+</html>
diff --git a/tests/test_to_html.py b/tests/test_to_html.py
@@ -1,13 +1,49 @@
-import os
+from pathlib import Path
 import filecmp
 import tempfile
 from src.cstag import to_html
 
 
 def test_html():
     cs = "cs:Z:=AC+GGG=T-ACGT*at~gt10cg=GNNN"
+    description = "Example"
+    cs_html = to_html(cs, description)
     output = tempfile.NamedTemporaryFile().name
+    Path(output).write_text(cs_html)
+    assert filecmp.cmp(output, Path("tests", "data", "to_html", "report.html"))
+
+def test_html_repeat_substitution():
+    cs = "cs:Z:=A*at*ag=A"
+    description = "Example"
+    cs_html = to_html(cs, description)
+    test = [h for h in cs_html.split("\n") if h.count("<p class='p_seq'>")]
+    test = test[0].split()
+    answer = Path("tests", "data", "to_html", "report_substitution.html").read_text().split("\n")
+    answer = [h for h in answer if h.count(r"<p class='p_seq'>")]
+    answer = answer[0].split()
+    assert test == answer
+
+
+def test_html_repeat_substitution_start():
+    cs = "cs:Z:*at*ag=A"
+    description = "Example"
+    cs_html = to_html(cs, description)
+    test = [h for h in cs_html.split("\n") if h.count("<p class='p_seq'>")]
+    test = test[0].split()
+    answer = Path("tests", "data", "to_html", "report_substitution_start.html").read_text().split("\n")
+    answer = [h for h in answer if h.count(r"<p class='p_seq'>")]
+    answer = answer[0].split()
+    assert test == answer
+
+
+def test_html_repeat_substitution_end():
+    cs = "cs:Z:=A*at*ag"
     description = "Example"
-    to_html(cs, output, description)
+    cs_html = to_html(cs, description)
+    test = [h for h in cs_html.split("\n") if h.count("<p class='p_seq'>")]
+    test = test[0].split()
+    answer = Path("tests", "data", "to_html", "report_substitution_end.html").read_text().split("\n")
+    answer = [h for h in answer if h.count(r"<p class='p_seq'>")]
+    answer = answer[0].split()
+    assert test == answer
 
-    assert filecmp.cmp(output + ".html", os.path.join("tests", "data", "to_html", "report.html"))