diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index 39c6e37..10679ab 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -54,6 +54,23 @@ jobs:
           directory: .
           snakefile: workflow/Snakefile
           args: "--lint --configfile config/example_config.yaml --config skip_version_check=True"
+
+  Pytest:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ['3.10', '3.11', '3.12', '3.13', '3.14']
+    steps:
+      - uses: actions/checkout@v6
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install test dependencies
+        run: python -m pip install --upgrade pip -r requirements-test.txt
+      - name: Run pytest suite
+        run: python -m pytest -q tests
 # Testing:
 #   runs-on: ubuntu-latest
 #   needs:
diff --git a/pyproject.toml b/pyproject.toml
index 52c96bc..1050e93 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,2 +1,7 @@
 [tool.snakefmt]
 line_length = 127
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+python_files = ["test_*.py", "*_test.py"]
+addopts = "-ra"
diff --git a/requirements-test.txt b/requirements-test.txt
new file mode 100644
index 0000000..42b555e
--- /dev/null
+++ b/requirements-test.txt
@@ -0,0 +1,4 @@
+pytest
+click
+numpy
+pandas
diff --git a/tests/README.md b/tests/README.md
new file mode 100644
index 0000000..161fdb3
--- /dev/null
+++ b/tests/README.md
@@ -0,0 +1,20 @@
+# Tests
+
+This directory contains the pytest suite for workflow scripts and supporting fixtures.
+
+## Layout
+
+- `tests/conftest.py` keeps shared pytest fixtures and makes the repo root importable.
+- `tests/<area>/test_*.py` holds the actual tests, grouped by script area such as `count`.
+- `tests/fixtures/<area>/` stores reusable input data for those tests.
+
+## Adding a new script test
+
+1. Put the new test in the matching area folder, for example `tests/count/test_new_script.py`.
+2. Add any reusable inputs under `tests/fixtures/<area>/`.
+3. Prefer `click.testing.CliRunner` for Click commands and pytest fixtures like `tmp_path` for temporary outputs.
+4. Run a focused check with `conda run -n mpralib python -m pytest -q tests/<area>/test_new_script.py`.
+
+## Current example
+
+The MPRAnalyze compiler test lives in `tests/count/test_mpranalyze_compiler.py` and uses the fixture input at `tests/fixtures/count/minimal_test_input.tsv`.
diff --git a/tests/conftest.py b/tests/conftest.py
new file mode 100644
index 0000000..0496f1d
--- /dev/null
+++ b/tests/conftest.py
@@ -0,0 +1,42 @@
+import sys
+from pathlib import Path
+
+import pytest
+from click.testing import CliRunner
+
+PROJECT_ROOT = Path(__file__).resolve().parents[1]
+PROJECT_ROOT_STR = str(PROJECT_ROOT)
+TESTS_ROOT = Path(__file__).resolve().parent
+
+if PROJECT_ROOT_STR not in sys.path:
+    sys.path.insert(0, PROJECT_ROOT_STR)
+
+
+@pytest.fixture(scope="session")
+def tests_root() -> Path:
+    return TESTS_ROOT
+
+
+@pytest.fixture(scope="session")
+def fixtures_root(tests_root: Path) -> Path:
+    return tests_root / "fixtures"
+
+
+@pytest.fixture(scope="session")
+def count_fixtures_root(fixtures_root: Path) -> Path:
+    return fixtures_root / "count"
+
+
+@pytest.fixture
+def minimal_count_input(count_fixtures_root: Path) -> Path:
+    return count_fixtures_root / "minimal_test_input.tsv"
+
+
+@pytest.fixture
+def ragged_missing_input(count_fixtures_root: Path) -> Path:
+    return count_fixtures_root / "ragged_missing_input.tsv"
+
+
+@pytest.fixture
+def cli_runner() -> CliRunner:
+    return CliRunner()
diff --git a/tests/count/test_mpranalyze_compiler.py b/tests/count/test_mpranalyze_compiler.py
new file mode 100644
index 0000000..c8f0f8d
--- /dev/null
+++ b/tests/count/test_mpranalyze_compiler.py
@@ -0,0 +1,140 @@
+import pandas as pd
+
+from workflow.scripts.count import mpranalyze_compiler as compiler
+
+
+class TestMpranalyzeCompiler:
+    def test_get_annot_parses_dna_and_rna_headers(self):
+        assert compiler.get_annot("DNA(condition X, replicate 1)") == ("DNA", "X", "1")
+        assert compiler.get_annot("RNA(condition Y, replicate 2)") == ("RNA", "Y", "2")
+
+    def test_get_annot_returns_none_for_unmatched_headers(self):
+        assert compiler.get_annot("Sequence") == (None, None, None)
+        assert compiler.get_annot("label") == (None, None, None)
+
+    def test_generate_annotation_output_repeats_and_numbers_barcodes(self):
+        input_frame = pd.DataFrame(
+            [
+                {"type": "DNA", "condition": "X", "replicate": "1"},
+                {"type": "RNA", "condition": "X", "replicate": "1"},
+            ]
+        )
+
+        output = compiler.generate_annotation_output(input_frame, number_barcodes=2)
+
+        assert list(output["sample"]) == ["DNA_X_1_1", "DNA_X_1_2", "RNA_X_1_1", "RNA_X_1_2"]
+        assert list(output["barcode"]) == ["1", "2", "1", "2"]
+
+    def test_generate_count_output_pads_and_flattens_by_barcode(self):
+        input_frame = pd.DataFrame(
+            [
+                {"label": "oligoA", "bc1": 10, "bc2": 20},
+                {"label": "oligoA", "bc1": 11, "bc2": 21},
+                {"label": "oligoB", "bc1": 12, "bc2": 22},
+            ]
+        ).set_index("label")
+
+        output = compiler.generate_count_output(input_frame, ["sample_1", "sample_2", "sample_3", "sample_4"], number_barcodes=2)
+
+        assert list(output["seq_id"]) == ["oligoA", "oligoB"]
+        assert list(output.loc[output["seq_id"] == "oligoA", ["sample_1", "sample_2", "sample_3", "sample_4"]].iloc[0]) == [10, 11, 20, 21]
+        assert list(output.loc[output["seq_id"] == "oligoB", ["sample_1", "sample_2", "sample_3", "sample_4"]].iloc[0]) == [12, 0, 22, 0]
+
+    def test_cli_generates_expected_count_tables(self, cli_runner, minimal_count_input, tmp_path):
+        rna_counts_output = tmp_path / "rna_counts.tsv.gz"
+        dna_counts_output = tmp_path / "dna_counts.tsv.gz"
+        rna_annotation_output = tmp_path / "rna_annot.tsv.gz"
+        dna_annotation_output = tmp_path / "dna_annot.tsv.gz"
+
+        result = cli_runner.invoke(
+            compiler.cli,
+            [
+                "--input",
+                str(minimal_count_input),
+                "--rna-counts-output",
+                str(rna_counts_output),
+                "--dna-counts-output",
+                str(dna_counts_output),
+                "--rna-annotation-output",
+                str(rna_annotation_output),
+                "--dna-annotation-output",
+                str(dna_annotation_output),
+            ],
+        )
+
+        assert result.exit_code == 0, result.output
+
+        rna: pd.DataFrame = pd.read_csv(rna_counts_output, sep="\t").set_index("seq_id")
+        dna: pd.DataFrame = pd.read_csv(dna_counts_output, sep="\t").set_index("seq_id")
+        cols= ["RNA_X_1_1", "RNA_X_1_2", "RNA_X_2_1", "RNA_X_2_2", "RNA_X_3_1", "RNA_X_3_2"]
+        row = rna.loc["oligoA"]
+        assert list(row.loc[cols]) == [
+            100,
+            101,
+            200,
+            201,
+            300,
+            301,
+        ]
+        cols= ["DNA_X_1_1", "DNA_X_1_2", "DNA_X_2_1", "DNA_X_2_2", "DNA_X_3_1", "DNA_X_3_2"]
+        row = dna.loc["oligoA"]
+        assert list(row.loc[cols]) == [
+            10,
+            11,
+            20,
+            21,
+            30,
+            31,
+        ]
+
+    def test_cli_missing_count_does_not_shift_replicates(self, cli_runner, ragged_missing_input, tmp_path):
+        # Regression guard for the MPRAflow-style "ragged within an oligo" bug
+        # (shendurelab/MPRAflow#87): a missing trailing count (empty cell / trailing
+        # tab) must stay a zero in its own replicate/barcode slot and must NOT shift
+        # later counts into the wrong replicate. The fixture's oligoA is missing its
+        # RNA replicate-3 / barcode-2 value, and oligoB has a single barcode (so it
+        # also exercises the across-oligo padding case).
+        rna_counts_output = tmp_path / "rna_counts.tsv.gz"
+        dna_counts_output = tmp_path / "dna_counts.tsv.gz"
+        rna_annotation_output = tmp_path / "rna_annot.tsv.gz"
+        dna_annotation_output = tmp_path / "dna_annot.tsv.gz"
+
+        result = cli_runner.invoke(
+            compiler.cli,
+            [
+                "--input",
+                str(ragged_missing_input),
+                "--rna-counts-output",
+                str(rna_counts_output),
+                "--dna-counts-output",
+                str(dna_counts_output),
+                "--rna-annotation-output",
+                str(rna_annotation_output),
+                "--dna-annotation-output",
+                str(dna_annotation_output),
+            ],
+        )
+
+        assert result.exit_code == 0, result.output
+
+        rna: pd.DataFrame = pd.read_csv(rna_counts_output, sep="\t").set_index("seq_id")
+        dna: pd.DataFrame = pd.read_csv(dna_counts_output, sep="\t").set_index("seq_id")
+
+        rna_cols = [
+            "RNA_X_1_1", "RNA_X_1_2", "RNA_X_1_3",
+            "RNA_X_2_1", "RNA_X_2_2", "RNA_X_2_3",
+            "RNA_X_3_1", "RNA_X_3_2", "RNA_X_3_3",
+        ]
+        # oligoA: the hole is at replicate 3 / barcode 2 -> must be 0, and the real
+        # barcode-3 value (302) must stay in barcode 3, not shift up to barcode 2.
+        assert list(rna.loc["oligoA", rna_cols]) == [100, 101, 102, 200, 201, 202, 300, 0, 302]
+        # oligoB has one barcode; remaining barcode slots pad with zeros per replicate.
+        assert list(rna.loc["oligoB", rna_cols]) == [103, 0, 0, 203, 0, 0, 303, 0, 0]
+
+        # DNA has no missing values; confirm nothing shifted there either.
+        dna_cols = [
+            "DNA_X_1_1", "DNA_X_1_2", "DNA_X_1_3",
+            "DNA_X_2_1", "DNA_X_2_2", "DNA_X_2_3",
+            "DNA_X_3_1", "DNA_X_3_2", "DNA_X_3_3",
+        ]
+        assert list(dna.loc["oligoA", dna_cols]) == [10, 11, 12, 20, 21, 22, 30, 31, 32]
diff --git a/tests/fixtures/count/minimal_test_input.tsv b/tests/fixtures/count/minimal_test_input.tsv
new file mode 100644
index 0000000..1a330ce
--- /dev/null
+++ b/tests/fixtures/count/minimal_test_input.tsv
@@ -0,0 +1,5 @@
+label	Sequence	Barcode	DNA(condition X, replicate 1)	DNA(condition X, replicate 2)	DNA(condition X, replicate 3)	RNA(condition X, replicate 1)	RNA(condition X, replicate 2)	RNA(condition X, replicate 3)
+oligoA	A	BC0	10	20	30	100	200	300
+oligoA	A	BC1	11	21	31	101	201	301
+oligoB	B	BC2	12	22	32	102	202	302
+oligoB	B	BC3	13	23	33	103	203	303
diff --git a/tests/fixtures/count/ragged_missing_input.tsv b/tests/fixtures/count/ragged_missing_input.tsv
new file mode 100644
index 0000000..cd53b62
--- /dev/null
+++ b/tests/fixtures/count/ragged_missing_input.tsv
@@ -0,0 +1,5 @@
+label	Sequence	Barcode	DNA(condition X, replicate 1)	DNA(condition X, replicate 2)	DNA(condition X, replicate 3)	RNA(condition X, replicate 1)	RNA(condition X, replicate 2)	RNA(condition X, replicate 3)
+oligoA	A	BC0	10	20	30	100	200	300
+oligoA	A	BC1	11	21	31	101	201
+oligoA	A	BC2	12	22	32	102	202	302
+oligoB	B	BC3	13	23	33	103	203	303
diff --git a/workflow/scripts/count/mpranalyze_compiler.py b/workflow/scripts/count/mpranalyze_compiler.py
index b559c1f..e758108 100644
--- a/workflow/scripts/count/mpranalyze_compiler.py
+++ b/workflow/scripts/count/mpranalyze_compiler.py
@@ -6,6 +6,44 @@
 import numpy as np
 import pandas as pd
 
+ANNOT_PATTERN = re.compile(r"^([DR]NA).*\(condition (.*), replicate (.*)\)$")
+
+
+def get_annot(head: str) -> tuple[str|None, str|None, str|None]:
+    match = ANNOT_PATTERN.match(head)
+    if match is not None:
+        group1 = match.group(1)
+        group2 = match.group(2)
+        group3 = match.group(3)
+        return (group1, group2, group3)
+    return (None, None, None)
+
+
+def generate_annotation_output(data, number_barcodes):
+    data = data.loc[data.index.repeat(number_barcodes)].copy()
+    data['barcode'] = data.groupby(['type', 'condition', 'replicate']).cumcount() + 1
+    data['barcode'] = data['barcode'].astype(str)
+    data['sample'] = data[['type', 'condition', 'replicate', 'barcode']].agg('_'.join, axis=1)
+    return data[["sample", "type", "condition", "replicate", "barcode"]]
+
+
+def generate_count_output(data, columns, number_barcodes):
+    rows = []
+    seq_ids = []
+    for label, group in data.groupby('label', sort=False):
+        padded = np.zeros((number_barcodes, data.shape[1]), dtype=np.int64)
+        vals = group.values[:number_barcodes].astype(np.int64)
+        padded[:len(vals)] = vals
+        rows.append(padded.flatten(order='F'))
+        seq_ids.append(label)
+    counts = pd.DataFrame(rows, columns=columns)
+    counts.insert(0, 'seq_id', seq_ids)
+    return counts
+
+
+def write_table(data, file):
+    data.to_csv(file, index=False, sep='\t', compression='gzip')
+
 
 # options
 @click.command()
@@ -39,12 +77,6 @@
 
 def cli(input_file, rna_counts_output_file, dna_counts_output_file, rna_annotation_output_file, dna_annotation_output_file):
 
-    annot_pattern = re.compile(r"^([DR]NA).*\(condition (.*), replicate (.*)\)$")
-    def get_annot(head):
-        m = annot_pattern.match(head)
-        if m is not None:
-            return m.group(1,2,3)
-
     # read input
     df = pd.read_csv(input_file,sep="\t", header='infer')
 
@@ -61,44 +93,26 @@ def get_annot(head):
 
     # counts for observation
 
-    dna_df = df.iloc[:,2:(2+n_dna_obs)].applymap(np.int64)
-    rna_df = df.iloc[:,(2+n_dna_obs):].applymap(np.int64)
+    dna_df = df.iloc[:,2:(2+n_dna_obs)].astype(np.int64)
+    rna_df = df.iloc[:,(2+n_dna_obs):].astype(np.int64)
 
     ## generate output DNA/RNA annotations (type_condition_replicate_barcode)
     n_bc = df.groupby('label').Barcode.agg(len).max()
-    def generateAnnotationOutput(data, number_barcodes):
-        data = data.loc[data.index.repeat(number_barcodes)]
-        data['barcode'] = data.groupby(['type','condition','replicate']).cumcount() +1
-        data['barcode'] = data['barcode'].astype(str)
-        data['sample'] = data[['type','condition','replicate','barcode']].agg('_'.join,axis=1)
-        data = data[["sample", "type", "condition", "replicate", "barcode"]]
-        return(data)
-    dna_annot = generateAnnotationOutput(dna_annot, n_bc)
-    rna_annot = generateAnnotationOutput(rna_annot, n_bc)
+    dna_annot = generate_annotation_output(dna_annot, n_bc)
+    rna_annot = generate_annotation_output(rna_annot, n_bc)
 
     ## generate output DNA/RNA count tables
     ## rows oligo/seq ids,/assignment then per barcode the counts. padding with zeros
-    def generateCountOutput(data,columns):
-        counts = pd.DataFrame(list(data.groupby('label').apply(lambda x: x.values.flatten()))).fillna(0).astype(np.int64)
-        counts.columns = columns
-        counts['seq_id'] = data.index.unique()
-        counts = counts[(['seq_id'] + list(columns))]
-        return(counts)
-
-    dna_counts = generateCountOutput(dna_df,dna_annot['sample'])
-    rna_counts = generateCountOutput(rna_df,rna_annot['sample'])
-
-    ## write table function
-    def write(data,file):
-        data.to_csv(file, index=False,sep='\t', compression='gzip')
+    dna_counts = generate_count_output(dna_df, dna_annot['sample'], n_bc)
+    rna_counts = generate_count_output(rna_df, rna_annot['sample'], n_bc)
 
     ## write output DNA/RNA annotations
-    write(dna_annot,dna_annotation_output_file)
-    write(rna_annot,rna_annotation_output_file)
+    write_table(dna_annot, dna_annotation_output_file)
+    write_table(rna_annot, rna_annotation_output_file)
 
     ## write output DNA/RNA annotations
-    write(dna_counts,dna_counts_output_file)
-    write(rna_counts,rna_counts_output_file)
+    write_table(dna_counts, dna_counts_output_file)
+    write_table(rna_counts, rna_counts_output_file)
 
 if __name__ == '__main__':
     cli()