tsalign 2.1.0__tar.gz → 2.1.2__tar.gz

{tsalign-2.1.0 → tsalign-2.1.2}/Cargo.lock

@@ -417,7 +417,7 @@ checksum = "e6d5a32815ae3f33302d95fdcb2ce17862f8c65363dcfd29360480ba1001fc9c"
 
 [[package]]
 name = "generic_a_star"
-version = "2.1.0"
+version = "2.1.2"
 dependencies = [
  "binary-heap-plus",
  "compare",
@@ -548,7 +548,7 @@ checksum = "bbd2bcb4c963f2ddae06a2efc7e9f3591312473c50c6685e1f298068316e66fe"
 
 [[package]]
 name = "lib_ts_chainalign"
-version = "2.1.0"
+version = "2.1.2"
 dependencies = [
  "binary-heap-plus",
  "bincode",
@@ -568,7 +568,7 @@ dependencies = [
 
 [[package]]
 name = "lib_tsalign"
-version = "2.1.0"
+version = "2.1.2"
 dependencies = [
  "binary-heap-plus",
  "compact-genome",
@@ -589,7 +589,7 @@ dependencies = [
 
 [[package]]
 name = "lib_tsshow"
-version = "2.1.0"
+version = "2.1.2"
 dependencies = [
  "ena",
  "itertools",
@@ -804,7 +804,7 @@ dependencies = [
 
 [[package]]
 name = "py_lib_tsalign"
-version = "2.1.0"
+version = "2.1.2"
 dependencies = [
  "compact-genome",
  "lib_tsalign",
@@ -1008,7 +1008,7 @@ dependencies = [
 
 [[package]]
 name = "seed_chain"
-version = "2.1.0"
+version = "2.1.2"
 dependencies = [
  "compact-genome",
  "generic_a_star",
@@ -1360,7 +1360,7 @@ checksum = "2b939c825d0c4295ac520f7b479927c1e5b458c9107a9d6f5ba65bc867bb93a5"
 
 [[package]]
 name = "tsalign"
-version = "2.1.0"
+version = "2.1.2"
 dependencies = [
  "anyhow",
  "bincode",
@@ -1381,7 +1381,7 @@ dependencies = [
 
 [[package]]
 name = "tsalign-tests"
-version = "2.1.0"
+version = "2.1.2"
 dependencies = [
  "anyhow",
  "clap",

tsalign-2.1.2/PKG-INFO

@@ -0,0 +1,144 @@
+Metadata-Version: 2.4
+Name: tsalign
+Version: 2.1.2
+Classifier: Programming Language :: Rust
+Classifier: Programming Language :: Python :: Implementation :: CPython
+Classifier: Programming Language :: Python :: Implementation :: PyPy
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown; charset=UTF-8; variant=GFM
+
+# Python Bindings for [`lib_tsalign`](https://crates.io/crates/lib_tsalign)
+
+[![PyPI](https://img.shields.io/pypi/v/tsalign)](https://pypi.org/project/tsalign/)
+
+Python bindings for the template switch aligner.  Aligns two DNA sequences
+while detecting template switches — short-range translocations where a query
+region is copied from (or aligns to) a different location, possibly on the
+reverse complement strand.
+
+## Installation
+
+```bash
+pip install tsalign
+```
+
+## Quick start
+
+```python
+import tsalign
+
+result = tsalign.align("ACGTACGT", "ACGACGT")
+print(result.cigar())   # compact alignment string
+print(result.stats())   # cost, duration, node counts, …
+```
+
+## Aligner options
+
+Create an `Aligner` once and reuse it for many sequences:
+
+```python
+aligner = tsalign.Aligner(
+    min_length_strategy="preprocess_lookahead",  # default: "lookahead"
+    chaining_strategy="lower_bound",             # default: "none"
+    total_length_strategy="maximise",            # default: "maximise"
+    no_ts=False,                                 # set True for plain gap-affine
+)
+
+result = aligner.align("ACGTACGT", "ACGACGT")
+```
+
+## Custom cost configuration
+
+Costs are specified in `.tsa` format.  Use `sample_tsa_config/config.tsa`
+as a starting point and consult the [main repository README](../README.md#modifying-the-cost-function) for a description
+of each parameter.
+
+```python
+aligner = tsalign.Aligner(costs_file="sample_tsa_config/config.tsa")
+result = aligner.align("ACGTACGT", "ACGACGT")
+```
+
+You can also pass the cost string directly:
+
+```python
+with open("my_costs.tsa") as f:
+    cost_str = f.read()
+aligner = tsalign.Aligner(costs=cost_str)
+```
+
+## Restricting the alignment range
+
+Use `AlignmentRange` to align only a window of the input sequences:
+
+```python
+from tsalign import Aligner, AlignmentRange
+
+aligner = Aligner()
+result = aligner.align(
+    "NNNACGTACGTNNN",
+    "ACGACGT",
+    range=AlignmentRange(reference_start=3, reference_end=11),
+)
+print(result.cigar())
+```
+
+Individual start/limit keyword arguments are also accepted when `range` is
+not provided:
+
+```python
+result = aligner.align(
+    "NNNACGTACGTNNN",
+    "ACGACGT",
+    reference_start=3,
+    reference_limit=11,
+)
+```
+
+## Working with alignment operations
+
+`alignment.alignments()` returns a typed list of `(count, op)` pairs:
+
+```python
+from tsalign import align, TemplateSwitchEntranceOp, TemplateSwitchExitOp
+
+result = align(reference, query)
+for count, op in result.alignments():
+    if isinstance(op, TemplateSwitchEntranceOp):
+        print(f"Template switch: {op.direction}, primary={op.primary}, offset={op.first_offset}")
+    elif isinstance(op, TemplateSwitchExitOp):
+        print(f"Exit, anti-primary gap: {op.anti_primary_gap}")
+    else:
+        # SimpleAlignmentOp — a basic edit in the primary or secondary track
+        print(f"{count}x {op.kind}")
+```
+
+## Visualisation
+
+```python
+result = tsalign.align(reference, query)
+result.viz_template_switches()   # prints ASCII art to stdout
+```
+
+## Limiting search resources
+
+```python
+result = aligner.align(
+    reference,
+    query,
+    cost_limit=100,       # return None if cost would exceed this
+    memory_limit=500_000, # return None if memory exceeds this number of bytes
+)
+if result is None:
+    print("No alignment found within limits")
+```
+
+## Accepted sequence types
+
+Any object whose `str()` representation is a valid DNA string (ACGTN) is
+accepted — including `Bio.Seq`:
+
+```python
+from Bio.Seq import Seq
+result = tsalign.align(Seq("ACGTACGT"), Seq("ACGACGT"))
+```
+

{tsalign-2.1.0/python_bindings → tsalign-2.1.2}/README.md

@@ -41,7 +41,7 @@ result = aligner.align("ACGTACGT", "ACGACGT")
 ## Custom cost configuration
 
 Costs are specified in `.tsa` format.  Use `sample_tsa_config/config.tsa`
-as a starting point and consult the main repository README for a description
+as a starting point and consult the [main repository README](../README.md#modifying-the-cost-function) for a description
 of each parameter.
 
 ```python

{tsalign-2.1.0 → tsalign-2.1.2}/generic_a_star/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "generic_a_star"
 description = "A generic implementation of the A* algorithm"
-version = "2.1.0"
+version = "2.1.2"
 edition.workspace = true
 rust-version.workspace = true
 license.workspace = true

{tsalign-2.1.0 → tsalign-2.1.2}/lib_tsalign/Cargo.toml

@@ -2,7 +2,7 @@
 name = "lib_tsalign"
 description = "A sequence-to-sequence aligner that accounts for template switches"
 authors = ["Sebastian Schmidt <sebastian.schmidt@helsinki.fi>"]
-version = "2.1.0"
+version = "2.1.2"
 license.workspace = true
 edition.workspace = true
 rust-version.workspace = true
@@ -26,9 +26,9 @@ thiserror = "2.0.3"
 num-traits.workspace = true
 serde = { workspace = true, features = ["derive"], optional = true }
 noisy_float = { version = "0.2.0" }
-generic_a_star = { version = "2.1.0", path = "../generic_a_star" }
+generic_a_star = { version = "2.1.2", path = "../generic_a_star" }
 log.workspace = true
 rustc-hash = "2.1.1"
-seed_chain = { version = "2.1.0", path = "../seed_chain" }
+seed_chain = { version = "2.1.2", path = "../seed_chain" }
 extend_map = "0.14.3"
 template-switch-error-free-inners = "0.1.1"

{tsalign-2.1.0 → tsalign-2.1.2}/lib_tsshow/Cargo.toml

@@ -3,13 +3,13 @@ name = "lib_tsshow"
 description = "A visualiser for template-switch alignments"
 license.workspace = true
 authors = ["Sebastian Schmidt <sebastian.schmidt@helsinki.fi>"]
-version = "2.1.0"
+version = "2.1.2"
 edition.workspace = true
 rust-version.workspace = true
 repository.workspace = true
 
 [dependencies]
-lib_tsalign = { version = "2.1.0", path = "../lib_tsalign" }
+lib_tsalign = { version = "2.1.2", path = "../lib_tsalign" }
 log.workspace = true
 svg = "0.18.0"
 resvg = "0.45.0"

{tsalign-2.1.0 → tsalign-2.1.2}/lib_tsshow/src/ts_arrangement/inner.rs

@@ -330,15 +330,19 @@ impl TsInnerArrangement {
                         .iter()
                         .take_while(|(_, c)| c.is_blank())
                         .last()
-                        .unwrap()
-                        .0;
+                        .map(|(i, _)| i);
                     let first_final_blank = inner
                         .iter()
                         .rev()
                         .take_while(|(_, c)| c.is_blank())
                         .last()
-                        .unwrap()
-                        .0;
+                        .map(|(i, _)| i)
+                        .unwrap_or(inner.len().into());
+                    let first_non_blank = inner
+                        .iter()
+                        .find(|(_, c)| !c.is_blank())
+                        .map(|(i, _)| i)
+                        .unwrap();
                     let first_source_column = inner
                         .iter_values()
                         .filter(|c| c.is_source_char())
@@ -355,7 +359,8 @@ impl TsInnerArrangement {
                     assert!(first_source_column >= last_source_column);
 
                     // Add prefix to extend to max_end.
-                    let mut arrangement_column = last_initial_blank + 1usize;
+                    let mut arrangement_column =
+                        last_initial_blank.map(|i| i + 1usize).unwrap_or(0.into());
                     let mut source_column = first_source_column;
                     #[allow(clippy::explicit_counter_loop)]
                     for _ in 0..ts.equal_cost_range.max_end {
@@ -384,15 +389,14 @@ impl TsInnerArrangement {
                     }
 
                     // Convert prefix to extend to min_end.
-                    let mut arrangement_column = last_initial_blank;
+                    let mut arrangement_column = first_non_blank;
                     for _ in 0..-ts.equal_cost_range.min_end {
-                        arrangement_column += 1;
-
                         while !inner[arrangement_column].is_source_char() {
                             arrangement_column += 1;
                         }
 
                         inner[arrangement_column].to_optional();
+                        arrangement_column += 1;
                     }
 
                     // Convert suffix to extend to max_start.

{tsalign-2.1.0 → tsalign-2.1.2}/pyproject.toml

@@ -10,6 +10,7 @@ classifiers = [
     "Programming Language :: Python :: Implementation :: CPython",
     "Programming Language :: Python :: Implementation :: PyPy",
 ]
+readme = "README.md"
 dynamic = ["version"]
 [tool.maturin]
 features = ["pyo3/extension-module"]

{tsalign-2.1.0 → tsalign-2.1.2}/python_bindings/Cargo.toml

@@ -6,7 +6,7 @@ authors = [
     "Sebastian Schmidt <sebastian.schmidt@helsinki.fi>",
     "Jasper Krauter <jasper.krauter@helsinki.fi>",
 ]
-version = "2.1.0"
+version = "2.1.2"
 edition.workspace = true
 rust-version.workspace = true
 repository.workspace = true
@@ -19,10 +19,10 @@ crate-type = ["cdylib"]
 
 [dependencies]
 pyo3 = "0.26"
-lib_tsalign = { version = "2.1.0", path = "../lib_tsalign", features = [
+lib_tsalign = { version = "2.1.2", path = "../lib_tsalign", features = [
     "serde",
 ] }
-lib_tsshow = { version = "2.1.0", path = "../lib_tsshow" }
+lib_tsshow = { version = "2.1.2", path = "../lib_tsshow" }
 serde.workspace = true
 pythonize = "0.26"
 pyo3-log = "0.13"

tsalign-2.1.2/python_bindings/README.md

@@ -0,0 +1,134 @@
+# Python Bindings for [`lib_tsalign`](https://crates.io/crates/lib_tsalign)
+
+[![PyPI](https://img.shields.io/pypi/v/tsalign)](https://pypi.org/project/tsalign/)
+
+Python bindings for the template switch aligner.  Aligns two DNA sequences
+while detecting template switches — short-range translocations where a query
+region is copied from (or aligns to) a different location, possibly on the
+reverse complement strand.
+
+## Installation
+
+```bash
+pip install tsalign
+```
+
+## Quick start
+
+```python
+import tsalign
+
+result = tsalign.align("ACGTACGT", "ACGACGT")
+print(result.cigar())   # compact alignment string
+print(result.stats())   # cost, duration, node counts, …
+```
+
+## Aligner options
+
+Create an `Aligner` once and reuse it for many sequences:
+
+```python
+aligner = tsalign.Aligner(
+    min_length_strategy="preprocess_lookahead",  # default: "lookahead"
+    chaining_strategy="lower_bound",             # default: "none"
+    total_length_strategy="maximise",            # default: "maximise"
+    no_ts=False,                                 # set True for plain gap-affine
+)
+
+result = aligner.align("ACGTACGT", "ACGACGT")
+```
+
+## Custom cost configuration
+
+Costs are specified in `.tsa` format.  Use `sample_tsa_config/config.tsa`
+as a starting point and consult the [main repository README](../README.md#modifying-the-cost-function) for a description
+of each parameter.
+
+```python
+aligner = tsalign.Aligner(costs_file="sample_tsa_config/config.tsa")
+result = aligner.align("ACGTACGT", "ACGACGT")
+```
+
+You can also pass the cost string directly:
+
+```python
+with open("my_costs.tsa") as f:
+    cost_str = f.read()
+aligner = tsalign.Aligner(costs=cost_str)
+```
+
+## Restricting the alignment range
+
+Use `AlignmentRange` to align only a window of the input sequences:
+
+```python
+from tsalign import Aligner, AlignmentRange
+
+aligner = Aligner()
+result = aligner.align(
+    "NNNACGTACGTNNN",
+    "ACGACGT",
+    range=AlignmentRange(reference_start=3, reference_end=11),
+)
+print(result.cigar())
+```
+
+Individual start/limit keyword arguments are also accepted when `range` is
+not provided:
+
+```python
+result = aligner.align(
+    "NNNACGTACGTNNN",
+    "ACGACGT",
+    reference_start=3,
+    reference_limit=11,
+)
+```
+
+## Working with alignment operations
+
+`alignment.alignments()` returns a typed list of `(count, op)` pairs:
+
+```python
+from tsalign import align, TemplateSwitchEntranceOp, TemplateSwitchExitOp
+
+result = align(reference, query)
+for count, op in result.alignments():
+    if isinstance(op, TemplateSwitchEntranceOp):
+        print(f"Template switch: {op.direction}, primary={op.primary}, offset={op.first_offset}")
+    elif isinstance(op, TemplateSwitchExitOp):
+        print(f"Exit, anti-primary gap: {op.anti_primary_gap}")
+    else:
+        # SimpleAlignmentOp — a basic edit in the primary or secondary track
+        print(f"{count}x {op.kind}")
+```
+
+## Visualisation
+
+```python
+result = tsalign.align(reference, query)
+result.viz_template_switches()   # prints ASCII art to stdout
+```
+
+## Limiting search resources
+
+```python
+result = aligner.align(
+    reference,
+    query,
+    cost_limit=100,       # return None if cost would exceed this
+    memory_limit=500_000, # return None if memory exceeds this number of bytes
+)
+if result is None:
+    print("No alignment found within limits")
+```
+
+## Accepted sequence types
+
+Any object whose `str()` representation is a valid DNA string (ACGTN) is
+accepted — including `Bio.Seq`:
+
+```python
+from Bio.Seq import Seq
+result = tsalign.align(Seq("ACGTACGT"), Seq("ACGACGT"))
+```

{tsalign-2.1.0 → tsalign-2.1.2}/seed_chain/Cargo.toml

@@ -3,7 +3,7 @@ name = "seed_chain"
 description = "A seeding and generic chaining mechanism for sequence-to-sequence alignment"
 license.workspace = true
 authors = ["Sebastian Schmidt <sebastian.schmidt@helsinki.fi>"]
-version = "2.1.0"
+version = "2.1.2"
 edition.workspace = true
 rust-version.workspace = true
 repository.workspace = true
@@ -12,5 +12,5 @@ repository.workspace = true
 [dependencies]
 compact-genome.workspace = true
 log.workspace = true
-generic_a_star = { version = "2.1.0", path = "../generic_a_star" }
+generic_a_star = { version = "2.1.2", path = "../generic_a_star" }
 num-traits.workspace = true

tsalign-2.1.0/PKG-INFO

@@ -1,7 +0,0 @@
-Metadata-Version: 2.4
-Name: tsalign
-Version: 2.1.0
-Classifier: Programming Language :: Rust
-Classifier: Programming Language :: Python :: Implementation :: CPython
-Classifier: Programming Language :: Python :: Implementation :: PyPy
-Requires-Python: >=3.8