An alignment is not reusable until matrix, gaps, and reference are named
An alignment parameter record is a rerun recipe, not a pretty screenshot. Prefer writing algorithm, matrix or match scores, gap open and extend, and the exact reference version beside the file. Reject citing an alignment that only exists as an image in a slide.
Needleman–Wunsch versus Smith–Waterman still owns the two algorithms. How to align DNA sequences still owns the procedure. Sequence file handoffs still own identity of the file. This page only walks the parameters that make the comparison reproducible. It does not invent a Zetta scoring matrix. It does not choose the algorithm for you.
The Record Must Rerun the Comparison
The objective is a record a second person can use to produce the same gaps and mismatches. That record is not the picture of the alignment. The picture is an output. The parameters are the method. If the method is missing, the output is an anecdote. Reviewers who ask “which BLOSUM” are not being pedantic. They are asking whether the gaps they see are a property of the sequences or a property of an unnamed penalty.
| Field | What to write | Fail if omitted |
|---|---|---|
| Algorithm or mode | Global, local, or the named program and version | A local hit is treated as an end-to-end proof |
| Matrix or match/mismatch | BLOSUM or PAM class, or nucleotide scores | Two “alignments” are not the same comparison |
| Gap open and extend | Both numbers, not “defaults” | Large indels shatter or vanish without a cause |
| Reference version | Construct ID and version, or assembly accession | Yesterday’s map is scored as today’s clone |
Record Checks Before Anyone Cites the Alignment
- Lock the reference. Write construct ID and version, or the genome accession and coordinates. Checkpoint: the file header and the record show the same identifier.
- Name the mode. Write global or local, or the program and version that implies the mode. Checkpoint: a later reader can say whether an unmatched end was allowed.
- Write the matrix or match/mismatch scores. Do not write “default.” Checkpoint: a second person can set the same scores without guessing the GUI theme.
- Write gap open and gap extend. If you changed them after looking at the result, write that you changed them and why. Checkpoint: the record distinguishes exploration from the cited comparison.
- Store the alignment file with those four fields, not as a cropped PNG. Checkpoint: the file reopens and the written parameters still describe it.
Expected Result and Verification
Expected result: a second person, given only the record and the two sequences, reproduces the same gap placement at the region you cited. Verification: they rerun the named program with the named parameters and the cited mismatch still sits at the same coordinate. If they have to ask which reference version you used, the record failed. If they have to guess the gap penalty because you wrote “standard settings,” the record failed. The how-to-align sibling already told you to record parameter changes. This page is that record, not another procedure.
A worked fail is a clone-verification slide that shows a red mismatch and no matrix. The reviewer cannot tell a true SNP from a local alignment that dropped a repeat. Another fail is “aligned in the cloud tool” with no version. The tool updated. The gaps moved. The figure did not. Verification is the rerun, not the confidence of the original analyst.
BLAST-style reports hide the same gap. A default word size, a default gap cost, and an unnamed database build will reproduce only until the service changes defaults. Write the program version and the database or construct identifier, not “BLAST online.” If you switched from a global pairwise tool to a local search after the first figure looked ugly, write that you switched. The later reader needs the comparison you cited, not the comparison you started with. Pairwise clone checks and multiple-sequence trees are different jobs. This checklist does not become a tree-building tutorial. It only insists that whichever job you cite still names matrix, gaps, and reference. A screenshot of a tree is still a screenshot.
Store the Named Alignment After the Fields Exist
Write the fields before anyone stores “aligned” on a map. If the lab only needs a teaching screenshot, stop. A parameter page is not the subject. If the comparison will be reused, name an alignment surface after the fields exist. ZettaGene is one such surface: official product language includes sequence alignment. The named alignment may be stored there. Official simulation scope remains restriction, Gibson, and homologous alignment. Golden Gate is not a Zetta feature. Do not treat this checklist as a Golden Gate product. Zetta does not invent a scoring matrix.
If the group still cites screenshots, you do not have a method. You have a picture. Write the four fields. Then cite the file. A PDF of the alignment with no header block is the same fail in a more official costume. The later reader still cannot rerun the gaps. Put the four fields in the record even if the PDF looks archival.
Software alone does not complete 21 CFR Part 11. This walk does not issue a compliance badge. It only keeps the comparison rerunnable. A named alignment in a notebook is still not a Part 11 record unless the laboratory’s own procedure says so.
Frequently Asked Questions
Is a screenshot of an alignment enough documentation?
No. Algorithm, scores, gaps, and reference version must be written.
Does this page choose Needleman–Wunsch or Smith–Waterman?
No. That sibling owns the two algorithms. This page only records the choice.
Related Articles