Help
Troubleshooting
Grouped by what you are seeing rather than by what is wrong, because you do not know what is wrong yet.
Something is wrong with a sequence
- It imported with no features
- It came in as FASTA or plain text, which cannot carry annotations. Re-export it as GenBank from wherever it came from and import again. File formats.
- The length is not what it should be
- Compare against the published sequence. A plasmid short by a few hundred bases has had something deleted — often years ago, by someone else, without the file being renamed. Check which features are missing before you build on it.
- A digest gives the wrong number of fragments
- Check the topology in the status bar. Cutting a circular plasmid once gives one fragment; cutting a linear one gives two. An off-by-one fragment count is almost always a plasmid imported as linear.
- Features are all shifted by one
- A coordinate-convention mismatch from the exporting tool. Check one feature against the source record; if it is off by one, they all are, and the fix is at the import boundary rather than one feature at a time.
- The translation is nonsense
- Check the strand before anything else. Several standard expression vectors carry their cassette on the bottom strand, and translating the wrong one produces stop codons everywhere. Use Translate and set the strand and frame explicitly.
A construct does not look right
- Verification says the reading frame is broken
- Usually a junction that is one or two bases out. Open the diff, look at the actual bases either side of the join, and translate from the start codon through it. Reviewing a change.
- Something essential is reported twice
- A fragment was assembled twice, or a whole backbone was duplicated. Check the total length first — it is usually obvious once you look at the number.
- The promoter is there but does not drive anything
- Both elements exist but are not arranged so one transcribes the other — wrong order, wrong orientation, or something in between them. This is a real defect, not a detection quirk.
- It is bigger than it should be
- Backbone plus insert, minus whatever the strategy removed. If the result is roughly double, it has been assembled twice.
- It built something different from the plan
- Say so, and point at the difference. Then restore the previous revision so the next attempt starts from a construct you trust.
The assistant is not helping
- It says it cannot edit files
- You are in Plan mode. That is what Plan mode is. Switch to Act, or approve a plan. Plan and Act.
- It keeps asking questions you have answered
- Answers live in one conversation. Put standing facts in
PLASMID.md, which is read at the start of every conversation in the project. Project instructions. - It used an enzyme I do not have
- It cannot know your freezer. List the enzymes you own in PLASMID.md and this stops happening.
- Answers have got worse in a long conversation
- A thread that has wandered across several constructs carries all of it as context. Start a new one — they stay grouped by project.
- It gave me an accession that does not exist
- A real and known failure. Do not use it. Ask it to search for the part again and give you a record you can open. Checking a record.
- It agreed with me when I was wrong
- It has a tendency to fold under pushback. If you challenge an answer and it reverses immediately, ask it to justify the original — that gets you to the evidence rather than to whichever answer you pushed for.
The interface
- Everything is greyed out
- You are browsing as a guest. Analysis tools, export, version history and file operations all need an account.
- I cannot find Analytics, Notifications, Plate Designer or Team
- They have no button anywhere. Press ⌘K and search, or use their shortcuts. Shortcuts.
- The Inventory buttons do nothing
- Correct — sample creation and movement are not connected yet. Inventory.
- A settings toggle changed nothing
- Several are not wired up. Account and settings lists which.
- My automation never ran
- Automations are a preview: nothing is saved and nothing is scheduled. GeneLoops, which you launch yourself, do run.
- The layout is a mess
- Reset Layout, from the avatar menu.
- It is unusable on my phone
- The workspace needs a laptop. On a phone you get the chat only.
I have been cut off
The message says which limit you reached and when it resets. The short burst window recovers in hours; the longer allowance takes until its own reset. The bench tools — maps, digests, primers, enzyme search, import and export — keep working either way. Usage limits.
Still stuck
What to include
Four things get to an answer much faster:
The project name; roughly when it happened; what you expected versus what you got; and your user ID from Settings → Account.
Write to support@gene-loop.com or use the contact form.