CRISPRPhylo

Use CRISPRPhylo (/tree-builder) to infer strain relationships from CRISPR spacer data. You can also request separate array trees within repeat groups. The page displays the strain tree, supports interactive rooting, and exports Newick and tree images. Use ArrayPhyloViz to combine a chosen tree with arrays.

Prepare a processed spacer CSV

Use Processed CSV from CRISPRPrep, either transferred through its CRISPRPhylo link or uploaded manually. Each row is one spacer; keep the order within each array. All seven columns must be present, even when some cells are empty:

Folder,Seq_Name,Orientation,Spacer_Sequence,DR_Consensus,Reverse_Complement_Sequence,Reverse_Complement_DR

Field	Meaning / check
Folder	Strain ID. Use the same ID for all arrays from one strain and distinct IDs for different strains.
Seq_Name	Array ID, consistent across spacer rows from that array.
Orientation	For -, reverse or reversed, the input validation uses the reverse-complement spacer column.
Spacer_Sequence	Original spacer sequence; used for other orientations.
DR_Consensus	Array repeat sequence; preserve the detector-derived value.
Reverse_Complement_Sequence	Required usable spacer sequence for reverse rows. Blank values can make those rows unusable.
Reverse_Complement_DR	Reverse-complement repeat; CRISPRPrep supplies it for negative arrays.

At least two distinct nonempty strain IDs and two usable spacer rows are required. This minimum does not guarantee informative cross-strain evidence. In particular, CRISPRFormer cannot infer a strain tree if no repeat group contains arrays from two different strains. Numbered TXT, genome FASTA and an unprocessed CSV lacking required columns are not valid substitutes.

Select the method and settings

Use GraphMSA by default. For a small dataset, such as 20 strains or fewer, you can choose CRISPRFormer. For larger datasets, use GraphMSA. The 20-strain guideline is a practical recommendation, not an input limit.

Setting	Default	Recommendation
Tree Inference method	GraphMSA	Keep GraphMSA for routine analyses and larger datasets. Choose CRISPRFormer for small datasets, for example 20 strains or fewer.
Model profile (CRISPRFormer)	High-variation data	Start with the default. Select Conservative data or Mixed data when that profile better matches your dataset.
Repeat grouping distance	0	Nonnegative edit-distance tolerance; keep 0 initially. Its meaning differs by method, as explained below.
Also build array trees	Unchecked	Enable before submission if you also need separate array trees, downloaded together as a ZIP.

The Repeat grouping distance label is broader than its current implementation. For CRISPRFormer, it sets the Levenshtein edit-distance threshold for repeat grouping (substitutions, insertions and deletions). For GraphMSA, the web adapter passes it as the spacer Levenshtein-distance tolerance within exact repeat families; increasing it does not merge different repeat families. Record both the method and value, and do not interpret an identical nonzero value as an equivalent setting across methods.

Submit and recover a run

1. Select the method, repeat-distance value and optional model profile.
2. Check Also build array trees if needed. It cannot be added to a completed result without a new run.
3. Paste CSV or upload one file; use only one input method. Load example supplies trial data.
4. Select Commit. After a run the same submit control may read Build tree.
5. Wait for the Job ID, then save Copy results link. Processing continues after the tab closes.
6. Return through My Jobs or Retrieve previous results to inspect and download the result.

The inference process has a 15-minute limit; queue waiting is separate. Results expire seven days after the task ends. Keep the original CSV and a settings record for reruns. The saved result report retains method/profile, repeat distance and counts, but not the submitted CSV or full run log. A blank Run log in a restored result does not by itself indicate failure.

Read the strain tree

Leaves are strains. Check that leaf names correspond to the intended Folder values. The saved report's strain count and usable spacer-row count describe the validated input, not statistical support; spacer rows are occurrences rather than distinct spacer types. Check the actual leaves too, since the inference depends on available repeat-group evidence.

Branching shows inferred grouping. Vertical order is a drawing arrangement, so adjacent labels are not necessarily the closest pair. A CRISPR-based grouping alone does not establish transmission or a whole-genome relationship.

Lengths and rooting need separate interpretation. Numbers after colons in View Newick source are branch lengths. Positive supplied lengths guide horizontal distances in the drawing; otherwise the viewer uses topology spacing. Do not interpret the tree as elapsed time without independent calibration. The web workflow does not add a separate bootstrap-confidence assessment.

For example, ((strain_A:0.1,strain_B:0.1):0.2,strain_C:0.3); groups A and B together in that rooted representation. The numeric branch lengths are not support percentages.

Drag inside the viewer to pan and use Ctrl + scroll (Command + scroll on Mac) to zoom. Scrolling without a modifier scrolls the page. With the viewer focused, use + / - to zoom and Home to fit the complete tree. The initial tree is fitted automatically.

Choose a root

Control	Effect
Midpoint root	Places the root halfway along the longest leaf-to-leaf path. Requires known nonnegative lengths on every edge and at least one positive length.
Root on branch	Enters branch-selection mode. Click a branch at the desired position to place a root there. Keyboard activation of a focused branch selects its midpoint.
Restore original root	Returns to the original inferred tree supplied by the saved result.

Press Escape to leave branch-selection mode. Midpoint rooting can be unavailable for trees with missing, negative or all-zero branch lengths. Use branch rooting only when you have a justified interpretation, such as an appropriate outgroup; a chosen display root is not new evidence for ancestry.

Root changes update the displayed tree, Newick source and downloads in the current page. The saved job retains its original tree. Download the changed Newick before leaving if you need to preserve that root, and use that downloaded tree in ArrayPhyloViz. Downloading directly from My Jobs retrieves the original saved tree.

Read optional array trees

When requested, Array trees reports the generated count and offers Download array trees (.zip). There are no individual array-tree previews on this page; the main preview remains the strain tree.

ZIP member	How to use it
group_*.nwk	A tree of arrays within one repeat family/group. These leaves represent arrays, not one leaf per strain.
index.csv	Group identifier, array count, tree filename and status. skipped_single_array means the group contained only one array and has no tree.
labels.csv	Use both group_id and tree_label to recover the original array_uid and identify strain/array membership. Labels are unique within a group, not necessarily across all groups.
README.txt	Description of the archive and its label mapping.

Array trees reuse the distances from the same analysis. Different files are separate within-group comparisons, not branches of one combined array tree. A count of zero can mean no group had enough arrays; inspect index.csv before treating it as an export failure. The ZIP is a separate download and is not embedded in the JSON report. It remains available from restored results until job expiry.

Export and continue

Select Download .nwk for the currently displayed root. Choose SVG or PNG, then Download tree, for an image of the complete current tree independent of pan/zoom. PNG uses 2× rendering and is limited to 32 million pixels and 16,000 pixels per side; use SVG for larger trees.

Keep the input CSV, settings, original Newick, any rerooted Newick, and optional array-tree ZIP. Open ArrayPhyloViz, add the chosen Newick and corresponding SpacerMap array-map SVG or existing array-per-line TXT, and review mapping: tree leaves use strain Folder, whereas SpacerMap CSV array names use Folder_Seq_Name when Folder is present. Assign by leaf handles prefixes without underscores; otherwise review mapping manually. SpacerMap's ID–sequence mapping download is not array-per-line TXT. Multiple arrays may belong to one strain.

Troubleshooting

Symptom	Action
Missing columns or too few valid rows	Use CRISPRPrep Processed CSV; check strain IDs and orientation-specific sequence fields.
No cross-strain repeat group	Check repeat sequences and sample grouping. For CRISPRFormer, reconsider repeat tolerance only if biologically justified.
Missing code/model or no output tree	Save the Job ID/error and contact the site administrator.
Midpoint root disabled	Inspect Newick lengths; missing, negative or all-zero lengths do not meet the requirements.
A root change disappeared on reopening	Expected: the job stores the original tree. Use the rerooted Newick you downloaded.
Array-tree ZIP missing	Confirm the option was enabled before submission; reopen interactive results rather than the generic report download.