Add needletail (#52723)

* Add needletail

* Add test file

* Add test files

* Include test files in meta.yaml

* Include test files in meta.yaml

* Fix test file

* Build with pip

* Add cffi requirement

* Remove ARM builds

* Rename test.py to run_test.py

* remove obselete test commands

* copy orjson recipe

* Delete recipes/needletail/build.sh

* skip version constraint lint

* use correct key for skip and maintainer

* set bindings and features for maturin

* add ARM builds

---------

Co-authored-by: Michael Hall <[email protected]>
Co-authored-by: Luiz Irber <[email protected]>

recipes/needletail/meta.yaml

@@ -0,0 +1,62 @@
+{% set name = "needletail" %}
+{% set version = "0.6.1" %}
+{% set sha256 = "58c1e04fc706060192fa2669327d45ebad1ab99fb15f73f2e040e4f8b1d051d9" %}
+
+package:
+  name: {{ name|lower }}
+  version: {{ version }}
+
+source:
+  url: https://github.com/onecodex/{{ name }}/archive/refs/tags/v{{ version }}.tar.gz
+  sha256: {{ sha256 }}
+  patches:
+    - patches/01-maturin-pyo3.patch
+
+build:
+  number: 0
+  run_exports:
+    - {{ pin_subpackage(name|lower, max_pin="x.x") }}
+  script:
+    - {{ PYTHON }} -m pip install . --no-deps --no-build-isolation --disable-pip-version-check
+    - cargo-bundle-licenses --format yaml --output THIRDPARTY.yml
+
+requirements:
+  build:
+    - python                              # [build_platform != target_platform]
+    - cross-python_{{ target_platform }}  # [build_platform != target_platform]
+    - crossenv                            # [build_platform != target_platform]
+    - maturin >=1,<2                      # [build_platform != target_platform]
+    - {{ compiler("c") }}
+    - {{ compiler("rust") }}
+    - {{ stdlib("c") }}
+    - cargo-bundle-licenses
+  host:
+    - python
+    - pip
+    - maturin >=1,<2 
+  run:
+    - python
+
+test:
+  files:
+    - test.fa
+    - test.fq
+
+about:
+  home: https://github.com/onecodex/needletail
+  license: MIT
+  license_family: MIT
+  license_file: 
+    - LICENSE
+    - THIRDPARTY.yml
+  summary: Fast FASTX parsing in Python
+  dev_url: https://github.com/onecodex/needletail
+
+extra:
+  skip-lints:
+    - version_constraints_missing_whitespace  # see https://github.com/bioconda/bioconda-recipes/issues/51185
+  recipe-maintainers:
+    - apcamargo
+  additional-platforms:
+    - linux-aarch64
+    - osx-arm64

recipes/needletail/patches/01-maturin-pyo3.patch

@@ -0,0 +1,15 @@
+diff --git a/pyproject.toml b/pyproject.toml
+index 6339e16..ae6a18e 100644
+--- a/pyproject.toml
++++ b/pyproject.toml
+@@ -9,4 +9,8 @@ classifier = [
+     "Programming Language :: Python :: 3",
+     "License :: OSI Approved :: MIT License",
+     "Topic :: Scientific/Engineering :: Bio-Informatics",
+-]
+\ No newline at end of file
++]
++
++[tool.maturin]
++bindings = "pyo3"
++features = ["python"]

recipes/needletail/run_test.py

@@ -0,0 +1,102 @@
+import unittest
+
+from needletail import (
+    parse_fastx_file,
+    parse_fastx_string,
+    NeedletailError,
+    reverse_complement,
+    normalize_seq,
+)
+
+
+FASTA_FILE, FASTQ_FILE = "test.fa", "test.fq"
+
+
+class ParsingTestCase(unittest.TestCase):
+    def get_fasta_reader(self):
+        return parse_fastx_file(FASTA_FILE)
+
+    def get_fastq_reader(self):
+        return parse_fastx_file(FASTQ_FILE)
+
+    def test_can_parse_fasta_file(self):
+        for i, record in enumerate(self.get_fasta_reader()):
+            if i == 0:
+                self.assertEqual(record.id, "test")
+                self.assertEqual(record.seq, "AGCTGATCGA")
+                self.assertIsNone(record.qual)
+                record.normalize(iupac=False)
+                self.assertEqual(record.seq, "AGCTGATCGA")
+                self.assertTrue(record.is_fasta())
+            if i == 1:
+                self.assertEqual(record.id, "test2")
+                self.assertEqual(record.seq, "TAGC")
+                self.assertIsNone(record.qual)
+                record.normalize(iupac=False)
+                self.assertEqual(record.seq, "TAGC")
+                self.assertTrue(record.is_fasta())
+
+            self.assertTrue(i <= 1)
+
+    def test_can_parse_fastq_file(self):
+        for i, record in enumerate(self.get_fastq_reader()):
+            if i == 0:
+                self.assertEqual(record.id, "EAS54_6_R1_2_1_413_324")
+                self.assertEqual(record.seq, "CCCTTCTTGTCTTCAGCGTTTCTCC")
+                self.assertEqual(record.qual, ";;3;;;;;;;;;;;;7;;;;;;;88")
+                record.normalize(iupac=False)
+                self.assertEqual(record.seq, "CCCTTCTTGTCTTCAGCGTTTCTCC")
+                self.assertTrue(record.is_fastq())
+            if i == 1:
+                self.assertEqual(record.id, "EAS54_6_R1_2_1_540_792")
+                self.assertEqual(record.seq, "TTGGCAGGCCAAGGCCGATGGATCA")
+                self.assertEqual(record.qual, ";;;;;;;;;;;7;;;;;-;;;3;83")
+                record.normalize(iupac=False)
+                self.assertEqual(record.seq, "TTGGCAGGCCAAGGCCGATGGATCA")
+                self.assertTrue(record.is_fastq())
+
+            self.assertTrue(i <= 2)
+
+
+class ParsingStrTestCase(ParsingTestCase):
+    def get_fasta_reader(self):
+        with open(FASTA_FILE) as f:
+            content = f.read()
+            return parse_fastx_string(content)
+
+    def get_fastq_reader(self):
+        with open(FASTQ_FILE) as f:
+            content = f.read()
+            return parse_fastx_string(content)
+
+
+class MiscelleanousTestCase(unittest.TestCase):
+    def test_normalize_seq(self):
+        self.assertEqual(normalize_seq("ACGTU", iupac=False), "ACGTT")
+        self.assertEqual(normalize_seq("acgtu", iupac=False), "ACGTT")
+        self.assertEqual(normalize_seq("N.N-N~N N", iupac=False), "N-N-N-NN")
+        self.assertEqual(normalize_seq("BDHVRYSWKM", iupac=True), "BDHVRYSWKM")
+        self.assertEqual(normalize_seq("bdhvryswkm", iupac=True), "BDHVRYSWKM")
+
+    def test_reverse_complement(self):
+        self.assertEqual(reverse_complement("a"), "t")
+        self.assertEqual(reverse_complement("c"), "g")
+        self.assertEqual(reverse_complement("g"), "c")
+        self.assertEqual(reverse_complement("n"), "n")
+
+        self.assertEqual(reverse_complement("atcg"), "cgat")
+
+
+class ErroringTestCase(unittest.TestCase):
+    def test_file_not_found(self):
+        with self.assertRaises(NeedletailError):
+            parse_fastx_file("hey")
+
+    def test_invalid_record(self):
+        with self.assertRaises(NeedletailError):
+            for i, record in enumerate(parse_fastx_string("Not a valid file")):
+                print(i)
+
+
+if __name__ == "__main__":
+    unittest.main()

recipes/needletail/test.fa

@@ -0,0 +1,4 @@
+>test
+AGCTGATCGA
+>test2
+TAGC

recipes/needletail/test.fq

@@ -0,0 +1,12 @@
+@EAS54_6_R1_2_1_413_324
+CCCTTCTTGTCTTCAGCGTTTCTCC
++
+;;3;;;;;;;;;;;;7;;;;;;;88
+@EAS54_6_R1_2_1_540_792
+TTGGCAGGCCAAGGCCGATGGATCA
++
+;;;;;;;;;;;7;;;;;-;;;3;83
+@EAS54_6_R1_2_1_443_348
+GTTGCTTCTGGCGTGGGTGGGGGGG
++
+;;;;;;;;;;;9;7;;.7;393333