01 / Quick start
A prediction in four steps
- 1
Choose input
Select pasted FASTA, file upload, or NCBI/UniProt accessions.
- 2
Add sequences
Enter, upload, or retrieve the protein records to be analyzed.
- 3
Choose results
Select the classification depth and Fast or Sensitive model.
- 4
Submit and save
Start the job, retain its private link, and download completed results.
02 / Prepare input
Three supported input routes
Whichever route you choose, the prediction server ultimately validates and submits protein FASTA. Fetched sequences are shown in an editable preview before the job begins.
Paste FASTA
Paste one or more protein records directly. Every record needs a non-empty header beginning with > and a sequence on the following line.
Upload a file
Upload .fasta, .fa, .txt, .csv, or .tsv. FASTA is loaded directly; a plain accession list is fetched using the selected database.
Use accessions
Enter NCBI Protein or UniProt accessions separated by commas, spaces, semicolons, or new lines. Up to 50 can be fetched at once.
Valid rice example
>sp|Q01883|RAG1_ORYSJ OS=Oryza sativa MASNKVVFSVLLLVVLSVLAAAMATMADHHQ... >sp|P0C510|RBL_ORYSA OS=Oryza sativa MSPQTETKASVGFKAGVKDYKLTYYTPEY...
Validation checklist
- ✓ Non-empty header after each >
- ✓ At least one amino-acid residue per record
- ✓ No numbers, gap characters, or punctuation in sequences
- ✓ Protein rather than nucleotide input
03 / Prediction levels
Choose the required classification depth
The selected level is passed to the RSLpred2 command-line model. Higher levels request the corresponding downstream classification defined by the hierarchy.
Single vs dual
Routes each protein to single or dual localization.
Single localization
Assigns a single-localized protein to one of 10 supported compartments.
Dual localization
Assigns a dual-localized protein to one of 6 supported compartment pairs.
Membrane topology
Classifies predicted membrane proteins as single-pass or multi-pass.
04 / Model strategy
Fast or Sensitive
Fast
Uses dipeptide amino-acid composition features. Choose it for larger batches or when shorter turnaround is the priority.
DPCP feature vector
Sensitive
Uses tripeptide amino-acid composition features. Choose it for smaller batches when prediction sensitivity is the priority.
TPC feature vector
05 / Submit and monitor
Keep the private job link
Choose Run Prediction after checking the input, level, and model. If anti-bot verification is enabled on the deployment, complete it before submission.
Queued
The server accepted the input and is waiting for execution.
Running
The job is being processed; status is checked every three seconds.
Completed
The browser opens Results automatically and the output becomes downloadable.
Copy the results URL shown after submission. It contains a private token, can resume job monitoring, and should not be shared publicly.
06 / Results and downloads
Read each available output level
The Results workspace presents available Level I–IV outputs above a horizontally scrollable table. Column names come directly from the predictor output.
Inspect in the browser
Switch among available result levels and scroll wide tables horizontally without truncating columns.
Export results
Download an individual level or export all available output files together for downstream analysis.
07 / Troubleshooting
Common problems
“FASTA input must begin with a > header.”+
Add a header line before every protein sequence, for example >protein_1. Do not place sequence characters before the first header.
“FASTA contains invalid amino-acid characters.”+
Remove numbers, punctuation, gaps, and nucleotide-only formatting. The server accepts standard amino-acid symbols plus B, X, Z, J, U, O, and * only.
An accession cannot be fetched.+
Confirm that the identifier belongs to the selected source—NCBI Protein or UniProt—and remove version or punctuation errors. You can paste FASTA instead if the record is unavailable remotely.
The job appears to remain queued or running.+
Keep the page open or save the private results link. The browser checks status every few seconds and resumes monitoring a locally remembered active job after a refresh.
The private results link no longer works.+
Check that the complete URL, including its private token, was copied. Results are retained for 30 days by default and cannot be restored after expiry.
08 / Limits and privacy
Current server boundaries
| Constraint | Default | What it means |
|---|---|---|
| Sequences per job | 10,000 | The deployment can override this value. |
| Sequence length | 20,000 residues | Maximum for any individual record. |
| Total residues | 1,000,000 | Maximum combined length of one submitted job. |
| Accession fetch | 50 IDs | Maximum fetched from NCBI or UniProt at once. |
| Result retention | 30 days | Keep a local download for long-term storage. |
Job results require the unguessable token embedded in the private link.
Do not submit confidential or personally identifying information in FASTA headers.
Ready to begin?
Open the workspace and load the included demonstration sequences.