Computed on the authors' server. This page is a proxy to the DeepEffector web server at deepeffector.hainanu.edu.cn (Hainan University). The protein language model (ProtT5 / esm1b) cannot run on this site's hardware, so sequences are forwarded there and only the score table is returned here. Upstream limits each submission to 50 sequences, so submit large proteomes in batches.
Input is a protein set, not necessarily a secretome. Unlike EffectorP, DeepEffector was benchmarked directly on whole proteomes (the authors screened all 22,487 proteins of Fusarium oxysporum f. sp. cubense TR4), so you may submit the complete proteome. Scores are model probabilities at a fixed 0.5 threshold — treat them as a prioritisation aid, not as experimental evidence.

Protein sequences (FASTA; one or more)

Loading limits…
Parameters: modelDeepEffector-Fungi threshold0.5 (fixed upstream)
Submitting to the DeepEffector server …
The upstream server queues the job and returns a score table when done. Please do not close this page.

Prediction results

Run details 

    

  


  

    ⚛
    Submit fungal or oomycete proteins to score them with the DeepEffector protein language model.
  







  
Recommended workflow

  
    
    
  1. Pick the model. Use DeepEffector-Fungi for ascomycetes and basidiomycetes, DeepEffector-Oomycete for Phytophthora, Pythium and relatives. The two models were trained on different taxa and are not interchangeable.
    2. 
    
  2. Submit up to 50 proteins per run. Whole proteomes work, but the upstream server caps each job at 50 sequences — split a proteome into batches of 50 and merge the TSV exports.
    3. 
    
  3. Prioritise candidates. Keep proteins above the 0.5 threshold and overlay infection-stage RNA-seq expression (see the Expression Atlas) or homology to validated effectors. Cross-check with EffectorP 3.0 on the same secretome: agreement between two independent models is stronger evidence than either score alone.
    4. 
  






  
Citation & implementation

  

    Li T, Meng Q, Wang D, Hu F, Zhang R, Chi X, Li J, Wang T, Li H. DeepEffector: a protein language model-driven deep learning framework for effector prediction in fungi and oomycetes.
    Plant Physiology 2026;202(2):kiag691.
    doi:10.1093/plphys/kiag691
     ·  official web server
  

  

    Server-side implementation: this page forwards FASTA to the official DeepEffector API and parses the returned score table;
    no model runs on this site. DeepEffector is trained on fungal and oomycete effectors only — it is not applicable to
    bacterial, nematode or insect effectors. If the upstream server is unreachable, use EffectorP 3.0 instead.