Feature/arpip impl by junniest · Pull Request #211 · acg-team/phyne

junniest · 2026-11-04T13:56:05Z

Implementation of indel event reconstruction under PIP and refactoring.

ARPIP implementation:

Added the asr module to pip_model with a PIPIndels struct that implements AncestralSequenceReconstruction.

API Improvement:

Added the from_msa_with_ancestral and from_msa_with_ancestral_unchecked methods to the AncestralAlignment trait, to construct a PhyloInfo with ancestral sequences from an existing alignment and a set of ancestral sequences.
from_msa_with_ancestral includes default implementation with validation checks for matching alphabets and sequence lengths between MSA and ancestral sequences and matching node IDs.

Utility Functions and Refactoring:

Introduced the mapping_to_bits utility function, which converts a sequence mapping to a FixedBitSet (will be potentially refactored away when the Mapping type changes to FixedBitSet from Vec<Option<usize>>).
Refactored the alphabet method in the Alignment trait and its implementations to always return a reference to a static Alphabet.
Moved validation logic for ancestor IDs to a dedicated function (validate_ancestor_ids) and adjusted imports accordingly.
Added the warn_on_extinct_columns function to provide users with warnings when columns in the alignment go extinct in all leaf sequences, suggesting the use of remove_extinct_columns.

Initial version of the ARPIP algo with homolgy path recpntruction

Integration test for PIP ASR using data from the ARPIP paper. Checks that homology paths are reconstructed correctly.

Added function to transform Mapping to FixedBitSet to alignment mod

Fixed test scenarios for ARPIP as they were checking the wrong tree node ids

Refactored PIP ASR to separate struct to match e.g. `ParsimonyPresenceAbsence`

Refactor to have `AncestralSequenceReconstruction` trait impl first for PIP asr and helpers second

Rename deletion_probabilities_at_nodes -> deletion_probabilities

Correct implementation of ARPIP indel inference with proper insertion probability computation

Added tests for when the most likely insertion location neitger at first node (MRCA) or root

Changed `Alignment::alphabet()` to return static ref to the alphabet to allow creating an MASA from an MSA.

Split `validate_ids_with_ancestors` in two methods, `validate_taxa_ids` and `validate_ancestor_ids` to allow validating them independently.

Added method to create an MASA from an Alignment and provided ancestral sequences in a `Sequence` object (`from_msa_with_ancestral`). Added a function `warn_on_extinct_columns` to reduce duplication for the check.

Use `warn_on_extinct_columns` in AncestralAlignment creation from all sequences

Added tests for `MASA::from_msa_with_ancestral` including correct creation and different types of possible errors

Cleaned up the tests for pip asr intermediate (tmp) values

Use `from_msa_with_ancestral` in PIP ASR to reuse MSA sequences rather than recreate a full `Sequences` object and use `from_aligned_with_ancestral_unchecked`.

Split `from_msa_with_ancestral` into two methods, `from_msa_with_ancestral` and `from_msa_with_ancestral_unchecked`. `from_msa_with_ancestral` has a default impl with the necessary checks and calls `from_msa_with_ancestral_unchecked`.

Add doctests for `from_msa_with_ancestral` and `from_msa_with_ancestral_unchecked`

Rename `set_internal` to `compute_internal_probabilities` and `set_leaf` to `compute_leaf_probabilities` to distinguish between the upward pass and the downward pass through the tree.

Refactored internal character presence determination into separate method and include presence in the tmp storage

Refactor to extract deciding deletion location (node or both children) to a separate method

github-actions · 2026-11-04T13:56:58Z

Test Results

1 184 tests +55 1 183 ✅ +55 15m 35s ⏱️ - 12m 16s
6 suites + 1 1 💤 ± 0
1 files ± 0 0 ❌ ± 0

Results for commit 5eaf7e0. ± Comparison against base commit 71d7ce2.

This pull request removes 8 and adds 63 tests. Note that renamed tests count towards both.

src/alignment/mod.rs - alignment ‑ surviving_columns_mask (line 167)
src/alignment/mod.rs - alignment::MASA ‑ from_aligned (line 480)
src/alignment/mod.rs - alignment::MASA ‑ from_aligned_with_ancestral_unchecked (line 588)
src/alignment/mod.rs - alignment::MASA ‑ remove_extinct_columns (line 662)
src/alignment/mod.rs - alignment::MSA ‑ alphabet (line 257)
src/alignment/mod.rs - alignment::MSA ‑ from_aligned_unchecked (line 349)
src/alignment/mod.rs - alignment::MSA ‑ len (line 285)
src/alignment/mod.rs - alignment::MSA ‑ seq_count (line 313)

alignment::tests ‑ masa_from_msa_and_ancestral
alignment::tests ‑ masa_from_msa_and_ancestral_unaligned_error
alignment::tests ‑ masa_from_msa_and_ancestral_wrong_length_error
alignment::tests ‑ masa_from_msa_with_ancestral_extra_ids
alignment::tests ‑ masa_from_msa_with_ancestral_missing_seqs
alignment::tests ‑ masa_from_msa_with_ancestral_wrong_alphabet
alignment::tests ‑ masa_from_msa_with_ancestral_wrong_seq_ids
alignment::tests::test_mapping_to_bits ‑ case_1
alignment::tests::test_mapping_to_bits ‑ case_2
alignment::tests::test_mapping_to_bits ‑ case_3
…

♻️ This comment has been updated with latest results.

Copilot

🟡 Changes recommended

Unresolved critical defects can select incorrect ancestors, require quadratic memory, and admit invalid or ambiguous sequence IDs that distort alignment statistics and ancestral updates.

5 open issues

What changed in this PR

Adds PIP-based ancestral indel reconstruction to the existing ancestral sequence reconstruction interface, with alignment-construction APIs and ARPIP comparison tests.

Changes:

Introduces PIPIndels to infer ancestral character presence from insertion and deletion probabilities.
Adds checked and unchecked ancestral-alignment constructors, refactors ID validation, and standardizes static alphabet references.
Adds reconstruction tests and ARPIP reference data.

File	Description
`phyne/tests/pip_asr.rs`	Compares reconstructed ancestral mappings with ARPIP results.
`phyne/src/pip_model/tests.rs`	Adds a reconstruction integration test.
`phyne/src/pip_model/mod.rs`	Exports the ASR module and `PIPIndels`.
`phyne/src/pip_model/asr.rs`	Implements indel reconstruction and focused unit tests.
`phyne/src/phylo_info/phyloinfo_builder.rs`	Refactors leaf and ancestor ID validation with tests.
`phyne/src/alignment/tests.rs`	Tests mapping conversion and ancestral-alignment constructors.
`phyne/src/alignment/mod.rs`	Adds constructors and mapping utilities; refactors alphabet access and extinct-column warnings.
`phyne/data/pip/arpip/sim-0_new.newick`	Supplies the ARPIP comparison tree.
`phyne/data/pip/arpip/sim-0_msa_new.fasta`	Supplies the observed leaf alignment.
`phyne/data/pip/arpip/sim-0_inferred.fasta`	Supplies reference ancestral and leaf sequences.

_{Review effort: Balanced}

MattesMrzik · 2026-11-05T14:29:35Z

+                "ancestral sequences must have the same length as the provided alignment"
+            );
+        }
+        validate_ancestor_ids(&phylo.tree, &ancestral_seqs)?;


copilot added

phylo.tree.node_ids_are_unique()?;

This seems like a valid suggestion to me

Like with the validate_taxa_ids it think this sequences object could contain seqs with ids that are in the tree tips set. they would not be detected by validate_ancestor_ids, in from_msa_with_ancestral they are silently ignored. So not quite like the MSA::from_aligned which added the additional seqs to the .seqs anyway rather than silently ignoring it.

MattesMrzik · 2026-11-05T14:20:56Z

+    let tip_sequence_ids = HashSet::from_iter(sequences.into_iter().filter_map(|rec| {
+        let id = rec.id().to_string();
+        if !internal_ids.contains(&id) {
+            Some(id)
+        } else {


This is true:

let tree = tree!("((A2:1,B3:1)I1:1,(C5:1,D6:1)I4:1)I0:0;"); let sequences = Sequences::with_alphabet_unchecked( vec![ record_wo_desc!("A2", b"XXX--X"), record_wo_desc!("B3", b"X-X--X"), record_wo_desc!("C5", b"X---X-"), record_wo_desc!("D6", b"--XXX-"), record_wo_desc!("I4", b"--XXX-"), ], Alphabet::dna(), ); let msa = MSA::from_aligned(sequences, tree.clone()).unwrap(); for seq in msa.msa.seqs() { println!("{}: {}", seq.id(), String::from_utf8_lossy(seq.seq())); } for map in msa.msa.leaf_maps().values() { println!("leaf map: {map:?}"); }

prints

A2: XXXX B3: XXX C5: XX D6: XXX I4: XXX leaf map: [Some(0), None, None, None, Some(1), None] leaf map: [Some(0), Some(1), Some(2), None, None, Some(3)] leaf map: [Some(0), None, Some(1), None, None, Some(2)] leaf map: [None, None, Some(0), Some(1), Some(2), None]

Either add a comment that set sizes should still be check or go back to 3 separate methods:

one that checks seq ids = tree tip ids

one that checks seq ids = tree internal ids

one that check seqs ids = all tree ids

This way there might be some duplicate code, but its more clear what each method does and when to apply it and it would catch incorrect seqs set sizes automatically.

+        tmp.f[idx] = f_child1.clone().component_mul(&tmp.f[children[1]]);
+        tmp.f[idx] *= 1.0 - tmp.deletion_probs[idx];
+        tmp.p[idx] =
+            f_child1.component_mul(&tmp.f[children[1]]) * tmp.insert_and_survive_probs[idx];


MattesMrzik · 2026-11-05T13:07:36Z

+    /// Tree is consumed and stored together with the ancestral alignment, because the alignment is only
+    /// meaningful in the context of the tree that it was built from.


I think this comment could be reworked to fit better: Its not the tree that is consumed, but the PhyloInfo.

Suggested change

/// Tree is consumed and stored together with the ancestral alignment, because the alignment is only

/// meaningful in the context of the tree that it was built from.

/// PhyloInfo (i.e., Tree and MSA) is consumed, the ancestral alignment and corresponding tree are returned together,

///because the alignment is only meaningful in the context of the tree that it was built from.

Yes, definitely. I think I copied the initial comment from from_aligned_with_ancestral_unchecked, which only gets a tree. I will rewrite, but now I am also thinking if I should flip the order of arguments, since from_aligned_with_ancestral_unchecked takes all_seqs and tree, whereas from_msa_with_ancestral_unchecked takes a PhyloInfo and ancestral_seqs.

Or, potentially, should we flip the arguments for from_aligned_with_ancestral_unchecked? It makes more sense to have tree+msa first and sequences second for the from_msa_... method, but then it clashes with the sequences+tree for the other one in my mind. It did not matter while there was only one method, but now it bothers me...

i guess either
tree, seqs & phyloInfo, seqs
or
seqs, tree & seqs, PhyloInfo

MattesMrzik · 2026-11-05T14:20:56Z

+    let tip_sequence_ids = HashSet::from_iter(sequences.into_iter().filter_map(|rec| {
+        let id = rec.id().to_string();
+        if !internal_ids.contains(&id) {
+            Some(id)
+        } else {


This is true:

let tree = tree!("((A2:1,B3:1)I1:1,(C5:1,D6:1)I4:1)I0:0;"); let sequences = Sequences::with_alphabet_unchecked( vec![ record_wo_desc!("A2", b"XXX--X"), record_wo_desc!("B3", b"X-X--X"), record_wo_desc!("C5", b"X---X-"), record_wo_desc!("D6", b"--XXX-"), record_wo_desc!("I4", b"--XXX-"), ], Alphabet::dna(), ); let msa = MSA::from_aligned(sequences, tree.clone()).unwrap(); for seq in msa.msa.seqs() { println!("{}: {}", seq.id(), String::from_utf8_lossy(seq.seq())); } for map in msa.msa.leaf_maps().values() { println!("leaf map: {map:?}"); }

prints

A2: XXXX B3: XXX C5: XX D6: XXX I4: XXX leaf map: [Some(0), None, None, None, Some(1), None] leaf map: [Some(0), Some(1), Some(2), None, None, Some(3)] leaf map: [Some(0), None, Some(1), None, None, Some(2)] leaf map: [None, None, Some(0), Some(1), Some(2), None]

MattesMrzik · 2026-11-05T14:29:35Z

+                "ancestral sequences must have the same length as the provided alignment"
+            );
+        }
+        validate_ancestor_ids(&phylo.tree, &ancestral_seqs)?;


copilot added

phylo.tree.node_ids_are_unique()?;

This seems like a valid suggestion to me

MattesMrzik · 2026-11-05T14:31:50Z

+                "ancestral sequences must have the same length as the provided alignment"
+            );
+        }
+        validate_ancestor_ids(&phylo.tree, &ancestral_seqs)?;


Like with the validate_taxa_ids it think this sequences object could contain seqs with ids that are in the tree tips set. they would not be detected by validate_ancestor_ids, in from_msa_with_ancestral they are silently ignored. So not quite like the MSA::from_aligned which added the additional seqs to the .seqs anyway rather than silently ignoring it.

MattesMrzik · 2026-11-05T14:33:12Z

+    /// Tree is consumed and stored together with the ancestral alignment, because the alignment is only
+    /// meaningful in the context of the tree that it was built from.


same remark as the comment I added to: from_msa_with_ancestral

MattesMrzik · 2026-11-05T14:57:48Z

+        let ancestral_seqs =
+            Sequences::with_alphabet_unchecked(ancestral_records, phylo.msa.alphabet())
+                .into_gapless();


I guess in this method:

fn from_msa_with_ancestral_unchecked( phylo: PhyloInfo<impl Alignment>, ancestral_seqs: Sequences, ) -> PhyloInfo<Self> {

when we check that we have a correct number of sequences and that they match the internal node ids (this is currently not the case as seqs with leaf ids can be in the passed ancestral_seqs), then we dont need to create this vec of ancestral_records in the for loop but can straight up take the passed ancestral_seqs and save it to MASA. we just need to assert the seqs alphabet is the same as the msa alphabet.

MattesMrzik · 2026-11-05T15:13:38Z

+    // sim-0_msa_new.fasta as the input msa
+    // sim-0_new.newick as the input tree (changed node ids from "x" -> "i_x")
+    // sim-0_inferred.fasta is the result for comparison (concatenated alignment from input and ARPIP reconstruction)
+    // Note: this dataset has no insertions that are more likely to be higher than the MRCA.


Perhaps say that you have tests for the case that insertion is higher than MRCA, so users reading that here know that you also test for that case that is not covered here.

MattesMrzik · 2026-11-05T15:20:43Z

+    // Deletion probabilities for each node, indexed by node (single value per branch).
+    deletion_probs: Vec<f64>,


I dont know how its for you in vscode but for me its nice to have these as /// comments since then I can get the hover text on that variable (also at every place where its used) and read the comment without having to jump to this line.

MattesMrzik · 2026-11-05T15:23:02Z

+}
+
+#[derive(Debug, Clone, PartialEq)]
+pub(crate) struct PIPAsrInfo<Q: QMatrix> {


Perhaps link the paper where the diagram that you made is in

MattesMrzik · 2026-11-05T16:11:28Z

+        // the potential values for f and p are precomputed but can be reset by deletion and insertion processing
+        let f_child1 = tmp.f[children[0]].clone();
+        tmp.f[idx] = f_child1.clone().component_mul(&tmp.f[children[1]]);
+        tmp.f[idx] *= 1.0 - tmp.deletion_probs[idx];
+        tmp.p[idx] =
+            f_child1.component_mul(&tmp.f[children[1]]) * tmp.insert_and_survive_probs[idx];


By the comment you mean that those values:

insert_and_survive_probs: insert_and_survive_probabilities(model, info), deletion_probs: deletion_probabilities(model, info),

are precomputed?
Or you are doing the precomputing here and some time later these (f and p) values might get set to a different value?

I think yes, perhaps change comment to:

/// precompute potential values for f and p. They might be overwritten by deletion and insertion processing

so that it is clear that the precomputing happens here and was not "precomputing performed somewhere else and the place here is where they are overwritten".

MattesMrzik · 2026-11-05T16:27:33Z

+        let ancestral_seqs =
+            Sequences::with_alphabet_unchecked(ancestral_records, phylo.msa.alphabet());
+
+        AA::from_msa_with_ancestral(phylo, ancestral_seqs).unwrap()


perhaps also call the unchecked version here?

MattesMrzik · 2026-11-06T10:06:39Z

+    // Data from the ARPIP paper, javapip simulation with mu = 0.1
+    // sim-0_msa_new.fasta as the input msa
+    // sim-0_new.newick as the input tree (changed node ids from "x" -> "i_x")
+    // sim-0_inferred.fasta is the result for comparison (concatenated alignment from input and ARPIP reconstruction)
+    // Note: this dataset has no insertions that are more likely to be higher than the MRCA.


perhaps also add this to a readme in the data/pip/arpip/readme

MattesMrzik · 2026-11-06T10:15:15Z

+    }
+}
+
+#[cfg(test)]


Suggested change

#[cfg(test)]

#[cfg(test)]

#[cfg_attr(coverage, coverage(off))]

i guess this is also missing in:

src/parsimony/matrices/mod.rs:502

src/parsimony_presence_absence/mod.rs:127

src/alignment/sequences.rs:412

src/tkf_model/tkf91.rs:221

src/tkf_model/mod.rs:196 (tkf_numerical_tests)

MattesMrzik · 2026-11-06T10:28:27Z

+        phylo: PhyloInfo<impl Alignment>,
+        ancestral_seqs: Sequences,
+    ) -> Result<PhyloInfo<Self>> {
+        if !ancestral_seqs.alphabet().eq(phylo.msa.alphabet()) {


Suggested change

if !ancestral_seqs.alphabet().eq(phylo.msa.alphabet()) {

if ancestral_seqs.alphabet() != phylo.msa.alphabet() {

MattesMrzik · 2026-11-06T10:34:53Z

+                -(-pip.mu() * node.blen).exp_m1() / pip.mu()
+            }) * common_factor
+        })
+        .collect::<Vec<f64>>()


Suggested change

.collect::<Vec<f64>>()

.collect()

MattesMrzik · 2026-11-06T11:10:52Z

+    #[test]
+    fn insertion_at_i5() {
+        let tree =
+            tree!("(((L3:1.0,L4:1.0)I2:1.0,(L6:0.001,L7:1.0)I5:10000.0)I1:1.0,L8:1.0)I0:0.0;");


In this test and similar ones we dont need the branches to be that short and that long, more moderate lens are sufficient, but could also keep it like this.

Simplified sequence/node match checks with helper function `check_node_sequence_ids_match`. Include validation that internal node ids are actually set (not empty) for `validate_ids_with_ancestors` and `validate_ancestor_ids`

Added tests for `ensure_ancestral_ids_set` and `check_node_sequence_ids_match` helpers

junniest added 30 commits October 21, 2026 13:53

Initial version of ASR under PIP

495393d

Initial version of the ARPIP algo with homolgy path recpntruction

Test data from ARPIP

a8a9af1

Integration test for PIP ASR

7e087dd

Integration test for PIP ASR using data from the ARPIP paper. Checks that homology paths are reconstructed correctly.

Add function to transform Mapping to FixedBitSet

c3dd7db

Added function to transform Mapping to FixedBitSet to alignment mod

Add test for mapping_to_bits

a2ec851

Fix test scenarios for ARPIP

f492dd2

Fixed test scenarios for ARPIP as they were checking the wrong tree node ids

Refactor PIP ASR to separate struct

1dd34cb

Refactored PIP ASR to separate struct to match e.g. `ParsimonyPresenceAbsence`

Fix tests to use new PIP asr struct

0440a40

Tests for PIP asr potential_character_presence

3000417

Refactor to have trait impl first for PIP asr

ad8d04b

Refactor to have `AncestralSequenceReconstruction` trait impl first for PIP asr and helpers second

Add insert_and_survive probability computation

950520a

Rename deletion_probs_at_nodes -> deletion_probs

b5cda2d

Rename deletion_probabilities_at_nodes -> deletion_probabilities

Fix mapping_to_bits

eea16a2

Add comment to test with ARPIP data

76f1c08

Correct implementation of ARPIP indel inference

46a3c10

Correct implementation of ARPIP indel inference with proper insertion probability computation

Fix tests to new subfunctions

3986316

Tests for when insertion location not at first node or root

1ec5f9c

Added tests for when the most likely insertion location neitger at first node (MRCA) or root

Change Alignment::alphabet() to return static &

4fcad37

Changed `Alignment::alphabet()` to return static ref to the alphabet to allow creating an MASA from an MSA.

Split validate_ids_with_ancestors in two methods

1983d6c

Split `validate_ids_with_ancestors` in two methods, `validate_taxa_ids` and `validate_ancestor_ids` to allow validating them independently.

Tests for new validate_ids_with_ancestors impl

79c1e26

Tests for validate_taxa_ids

6a98098

Tests for validate_ancestor_ids

c162d67

Add missing use statement

324b3d7

Add method from_msa_with_ancestral for MASA

37a38b8

Added method to create an MASA from an Alignment and provided ancestral sequences in a `Sequence` object (`from_msa_with_ancestral`). Added a function `warn_on_extinct_columns` to reduce duplication for the check.

Use warn_on_extinct_columns in anc alignment

4f152a9

Use `warn_on_extinct_columns` in AncestralAlignment creation from all sequences

Add tests for MASA::from_msa_with_ancestral

7bdab34

Added tests for `MASA::from_msa_with_ancestral` including correct creation and different types of possible errors

Cleanup pip asr tmp value tests

a3162e6

Cleaned up the tests for pip asr intermediate (tmp) values

Use from_msa_with_ancestral in PIP ASR

43e2d8e

Use `from_msa_with_ancestral` in PIP ASR to reuse MSA sequences rather than recreate a full `Sequences` object and use `from_aligned_with_ancestral_unchecked`.

Split from_msa_with_ancestral to have unchecked

7111134

Split `from_msa_with_ancestral` into two methods, `from_msa_with_ancestral` and `from_msa_with_ancestral_unchecked`. `from_msa_with_ancestral` has a default impl with the necessary checks and calls `from_msa_with_ancestral_unchecked`.

Add doctests for from_msa_with_ancestral

2c08fd0

Add doctests for `from_msa_with_ancestral` and `from_msa_with_ancestral_unchecked`

junniest added 11 commits November 4, 2026 12:28

Rename set_internal and set_leaf

5ef5074

Rename `set_internal` to `compute_internal_probabilities` and `set_leaf` to `compute_leaf_probabilities` to distinguish between the upward pass and the downward pass through the tree.

Refactor internal character presence into method

6cd26d4

Refactored internal character presence determination into separate method and include presence in the tmp storage

Refactor to extract deciding deletion location

e7feab9

Refactor to extract deciding deletion location (node or both children) to a separate method

Extract insertion prob computation to method

1da72b6

Refactor to group methods better

dcb7b2c

Cleanup leaf computation

57581d0

Remove unnecessary clone

b7300d5

Renames for readability

1218a5f

Add comment to internal value computation

b3b9a8e

Fix comments

43bbcae

Refactor internal sequence creation

fd124ae

junniest requested review from MattesMrzik and a balanced review from Copilot November 4, 2026 13:56

Copilot started reviewing on behalf of junniest November 4, 2026 13:57

View session

Copilot AI reviewed Nov 4, 2026

View reviewed changes

junniest added 2 commits November 4, 2026 17:51

Simplify deletion set storage per column

d2660e2

Remove repeated calls to tree.leaf_ids()

ef86dfe

MattesMrzik reviewed Nov 5, 2026

View reviewed changes

MattesMrzik requested changes Nov 6, 2026

View reviewed changes

junniest added 5 commits November 6, 2026 13:55

Fix comment PhyloInfo::from_msa_with_ancestral

21322e5

Rework sequence/node match checks

217f702

Simplified sequence/node match checks with helper function `check_node_sequence_ids_match`. Include validation that internal node ids are actually set (not empty) for `validate_ids_with_ancestors` and `validate_ancestor_ids`

Fix alignment tests to new error messages

5f228f8

Fix PhyloInfo tests to new validation logic

29281f9

Add tests for helper functions

5eaf7e0

Added tests for `ensure_ancestral_ids_set` and `check_node_sequence_ids_match` helpers

		/// Tree is consumed and stored together with the ancestral alignment, because the alignment is only
		/// meaningful in the context of the tree that it was built from.

		// Deletion probabilities for each node, indexed by node (single value per branch).
		deletion_probs: Vec<f64>,

	#[cfg(test)]
	#[cfg(test)]
	#[cfg_attr(coverage, coverage(off))]

	if !ancestral_seqs.alphabet().eq(phylo.msa.alphabet()) {
	if ancestral_seqs.alphabet() != phylo.msa.alphabet() {

Conversation

junniest commented Nov 4, 2026

Uh oh!

github-actions Bot commented Nov 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Test Results

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

MattesMrzik Nov 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MattesMrzik Nov 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MattesMrzik Nov 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MattesMrzik Nov 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

github-actions Bot commented Nov 4, 2026 •

edited

Loading

MattesMrzik Nov 5, 2026 •

edited

Loading

MattesMrzik Nov 5, 2026 •

edited

Loading

MattesMrzik Nov 5, 2026 •

edited

Loading

MattesMrzik Nov 5, 2026 •

edited

Loading