Keep source maps when pruning trees and alignments
Understand why an isolated root, an unknown branch length or a plausible trimmed alignment can hide a different task; verify the complete source and edit map before handoff.
A plausible summary can answer a different task
A URDF model can have one candidate root and still contain an unreachable cycle. A pruned tree can retain the desired taxa but move the root. A filtered alignment can have attractive row lengths while applying different masks to different rows. Start with the declared task and retain the original source beside the result.
The three tools attach complete source text, SHA-256 provenance and coordinate spans to scoped reports. They do not execute the robot model, infer a biological tree or create an alignment.
Follow the isolated world link and the other component
The posted tetrahedral model motivates checking roots, weak components and directed cycles together. The example report’s four links, three joints, isolated world root and separate three-link cycle make validDeclaredTree=false. Returning only the first root would hide the relevant failure.
When a target is undeclared, the report keeps the declared name and its source. Unknown XML and mesh URLs remain inert and unreviewed; a graph diagnosis cannot substitute for a complete URDF or simulator review.
Known decimal distances and absent lengths need different handling
Removing B and C from ((A:0.1,B:0.2):0.3,C:0.4):0.7; yields (A:0.4):0.7; because both contracting edge lengths are present and their exact sum is0.4. The original root remains. With an absent A length, the unary node survives; guessing zero would change what the source says.
Internal support labels and comments prevent contraction when it would discard them. Review duplicate/unmatched requests and every contraction record. A request for selected taxa is not permission to reroot at their MRCA.
| Source feature | Required check |
|---|---|
| Original root | Root identity and root-edge token remain |
| Missing branch length | Unknown flag remains; no zero replacement |
| Label/comment on unary node | Annotation remains instead of being silently contracted |
| Requested taxa list | Duplicates and unmatched requests stay visible |
Apply one threshold and one map to the entire alignment
For10 rows at90%,9 gaps removes the column and8 gaps does not. The >= rule uses integer cross multiplication, avoiding a rounded fraction near the boundary. N/X are ordinary supported residues until explicitly selected as missing.
Use missingCounts and columnMap to reconstruct all projected rows. Keep full headers, duplicate row IDs and original UTF-8/newline spans. The empty-column result is explicit and valid; it is not a missing output. An input above10MiB, including the source question’s220MB dataset, needs another workflow rather than an untested size promise.
Handoff complete files and label previews accurately
The visible table is at most200 rows, with each long cell clipped after2000 UTF-16 units. Text/copy above20000 UTF-16 units is a compact preview. Complete JSON/CSV and Newick/FASTA downloads are checked against the combined20/20/40MiB budgets and published only as a whole.
Give the receiver original source-bearing JSON and every matching output file. Use the exact source spans and complete maps for verification, and treat graph issues or unsupported profiles as evidence to investigate in the producer. Closing the page loses browser memory; downloaded files are the lasting handoff.
- Check the full report and source hash, not only a copied preview.
- Safe CSV prefixes formula-looking text; use JSON to compare exact original names.
- Keep source and result paired when changing taxa, missing symbols or thresholds.
- A successful processing status means a scoped report was generated, not that the model or biological choice is correct.
References
- Tetrahedral robot simulation question
First-person demand: the posted model has an isolated world link and a separate three-link cycle. Only declared graph structure is reviewed here; simulator behavior is not reproduced.
- Rooted Newick species selection question
First-person demand for retaining a specified species list; a later comment explicitly says the tree is rooted. Keeping the original root is distinct from extracting an MRCA subtree.
- Whole-column FASTA gap filtering question
First-person request for whole-column filtering at at least90% gaps. The reported220MB dataset is outside this tool’s10MiB input budget.
- URDF model declarations
Correctness reference for declared links and joints, not demand evidence or a full URDF/simulation certificate.
- Newick tree format
Correctness reference. This declared profile preserves literal underscores and rejects anonymous leaves as Unsupported, without saying those variants are universally invalid.
- Biopython alignment IO
Independent-reader reference for aligned sequences; the trimming rule remains the explicitly chosen missing set and threshold.
Tools in this category
Expand a tool to see its steps, options and supported formats, then open its workspace.
URDF declared-link graph reviewInspect declared URDF links and joints for unresolved references, multiple parents, roots and cycles, preserving the full source and exact byte positions.
Explain a root-link error before opening a simulator. Review the whole declared parent-child graph, including isolated roots and separate cycles; keep the original XML beside every finding.
Steps
- Select or paste one already-expanded UTF-8 robot XML. Run the declared graph review.
- Inspect unresolved references, root count, separate components and cycle components together. Depth is available only for a valid declared tree.
- Download complete JSON/CSV and check byte locations against the original source before changing the model.
Available options
- Protect formula-like CSV text
- On by default
Capabilities and limits
- One already-expanded UTF-8 robot XML up to 5 MiB, with optional UTF-8 BOM; up to 10,000 links, 10,000 joints and XML depth 64. Complete source-containing JSON and CSV together must fit 20 MiB. All limits apply together.
- Audits declared link/joint names, parent-child references, unresolved links, multiple incoming joints, roots, strongly connected cycles, weak components and depths of a valid declared tree. A report can complete while genuine graph issues remain; one root alone is insufficient.
- Duplicate names, missing or repeated parent/child references and malformed XML reject the whole input. Unresolved declared links, self-loops, cycles and multiple parents appear as structural issues. Names are exact and case-sensitive.
- DTD/entity declarations, processing instructions, namespaced elements/xacro and unresolved ${...} or $(...) names are Unsupported. Built-in and numeric XML character references are decoded as inert names. No mesh, package, URL or external file is opened.
- Joint types, model versions, visual/collision/inertial/dynamics/mimic and unknown XML content remain unreviewed. Their original source is retained. This is not whole-URDF schema validation, physics, simulation, closed-loop repair or a claim that a robot is safe to run.
- Full JSON includes original UTF-8 source, filename/hash, all links/joints and their exact UTF-8 byte spans, each parent/child tag, components and issues. CSV contains every link/joint. Preview shows 200 rows and 2,000 UTF-16 units per cell; large reports copy only the labelled preview. Downloads remain complete.
Rooted Newick taxa pruningKeep or remove named taxa from a declared rooted Newick tree while retaining the original root, unknown edge lengths, annotations and exact decimal contractions.
Supply one rooted tree and a JSON list of exact leaf labels. Trace every kept, removed, missing or repeated request to the original tree; download the changed Newick with full provenance.
Steps
- Choose one rooted UTF-8 tree and confirm the rooted-tree declaration. Enter exact labels as a JSON array.
- Choose keep or remove; review kept/removed taxa, unmatched requests and exact edge changes.
- Download pruned Newick and full JSON/CSV. Check original-root paths and unknown-length states before using the result.
Available options
- I declare this tree rooted; retain original root
- On by default
- Request-list mode
- Keep listed taxa · Remove listed taxa
- Protect formula-like CSV text
- On by default
Capabilities and limits
- One UTF-8 single Newick tree up to 5 MiB, optional BOM, plus taxa JSON array up to 1 MiB. Up to 50,000 nodes, 10,000 internal-parenthesis depth and 10,000 requested taxa; each decoded label and numeric token up to 4,096 UTF-16 units; exponent absolute value up to 100. Complete JSON, CSV and Newick together must fit 20 MiB.
- Explicitly declare the tree rooted. Keep/remove use exact case-sensitive labels with literal underscores; quoted labels and doubled single quotes are supported. Duplicate source leaf labels are ambiguous and reject; duplicate requests and unmatched labels remain separately reported.
- The original root and its edge remain. Only an unlabeled, unannotated non-root unary node whose incoming and surviving-child lengths both exist is contracted; decimal sums are exact, including values outside float64. Missing lengths remain unknown and never become zero.
- Surviving internal labels/support and comments stay attached to their nodes. Unknown or annotated unary nodes remain. This is not MRCA extraction, re-rooting, unrooted minimum-subtree search, strict minimum-node optimization or biological inference.
- Anonymous leaves and unsupported leading directives, including [&U], are outside this profile; [&R] is retained. These restrictions do not redefine valid Newick. Removing every leaf rejects the batch; change the mode or request list to retain at least one leaf.
- Full JSON preserves original sources and hashes, all original node/length spans, every request including duplicates, kept/removed/unmatched taxa and exact contractions. CSV includes every source taxon and unmatched unique request. Preview shows 200 rows/2,000 UTF-16 units per cell; copying large reports gives only a labelled preview. Complete downloads are atomic.
Aligned FASTA whole-column trimmingRemove alignment columns using an explicit missing-character set and threshold, preserving every row, full header and complete original-to-output column map.
Apply one column decision to all equal-width sequences. Inspect missing counts and retained coordinates before downloading the projected alignment; ambiguous N/X characters count only when explicitly selected.
Steps
- Select or paste one equal-width aligned FASTA. Choose the explicit missing-character set and column rule.
- Run trimming and inspect original, retained and removed column counts. Check the complete column map when positions matter.
- Download projected FASTA and full JSON/CSV. Keep the original source and coordinates with the handoff.
Available options
- Explicit missing characters, no separator
- -
- Whole-column removal rule
- At least percentage · Any selected missing character
- Integer percentage threshold
- 90
- Protect formula-like CSV text
- On by default
Capabilities and limits
- One equal-width UTF-8 aligned FASTA up to 10 MiB, with optional BOM and LF/CRLF. Up to 10,000 rows, 100,000 source columns and 8,000,000 total residue cells; full headers up to 65,536 UTF-16 units and at most 200,000 nonempty physical sequence fragments. Complete JSON, CSV and FASTA together must fit 40 MiB.
- Residues are ASCII A–Z/a–z, *, ?, . or -. Full UTF-8 headers, duplicate IDs and source spans remain distinct. Lone CR, unsupported residues, unequal widths and control characters in headers reject; blank source lines are retained in the original text.
- Choose missing characters explicitly, then any missing value or an integer 0–100% rule. Percentage removes a column when missing count ×100 is greater than or equal to row count ×threshold. At90%,9 of10 removes and8 of10 remains; at0%, every column is removed.
- N/X ambiguity is not a gap unless you select it. This tool does not align sequences, infer relationships or choose biologically optimal trimming. The cited220MB dataset is outside the10MiB profile.
- Full JSON retains original source/hash, every full header, physical source spans, all missing counts, the complete zero-based column map and every projected sequence. CSV maps every source column; null/new empty coordinate means removed. No dense lists of millions of cause indices are silently truncated.
- Zero-column output is valid and explicit: every original header has an empty sequence. Output FASTA uses LF; original LF/CRLF/BOM remain reconstructable from report source text. Preview is200 rows/2,000 UTF-16 units per cell; large copies are labelled previews while complete downloads remain atomic.