fimep 0.1.3

An ensemble predictor that uses a deep neural network for effector prediction. Input is a dataframe of predictions from kingdom-specific programs.

fimep

fimep is a Python package with a command-line interface (CLI) that leverages a deep learning model to predict effectors. It integrates predictions from multiple effector prediction programs to provide improved accuracy in identifying pathogen effector proteins. The package supports predictions for fungi, bacteria, oomycete, and can handle all kingdoms simultaneously.

---

Installation

To use fimep, install it from PyPI using:

pip install fimep

Optional: users can create a virtual environment using mamba or python venv before installing fimep.

---

Supported tools and kingdoms

To run predictions with fimep, user must first generate outputs using the supported effector prediction programs listed below. Each program name links to its respective repository or documentation.

Kingdom	Supported tools
Fungi	EffectorP-3.0, deepredeff, WideEffHunter
Bacteria	deepredeff, EffectiveT3, T3SEpp
Oomycete	EffectorP-3.0, deepredeff, WideEffHunter, EffectorO
All	EffectorP-3.0, deepredeff, WideEffHunter, EffectorO, EffectiveT3, T3SEpp

---

Usage

fimep can be used in two ways:

1. Single workflow (runall): Complete analysis from raw program outputs to final predictions
2. Step-by-step: Individual steps for more control over the process

Single workflow usage

It is best practice to type in each command into the terminal directly.

Note

• WideEffHunter: Requires the complete input file and predicted output file from WideEffHunter program.

For fungi

fimep runall \
    --effectorp fungi_effp.txt \
    --deepredeff dr_result.csv \
    --wideeffhunter complete.fasta pred.fasta \
    --kingdom fungi \
    --output final_fungi_prediction.csv

For oomycetes

fimep runall \
    --effectorp ep_result.txt \
    --deepredeff dr_result.csv \
    --effectoro eo_result.csv \
    --wideeffhunter complete.fasta pred.fasta \
    --kingdom oomycete \
    --output final_oomycete_prediction.csv

For bacteria

fimep runall \
    --t3sepp t3sepp_result.txt \
    --deepredeff dr_result.csv \
    --effectivet3 et3_result.csv \
    --kingdom bacteria \
    --output final_bacteria_prediction.csv

---

Step-by-step usage

The usage include the subcommand and options which are explained below:

fimep <subcommand> [options]

Available subcommands:

Command	Description
preprocess_effectorp	Format raw EffectorP-3.0 output into standard structure
preprocess_effectiveT3	Format raw EffectiveT3 output
preprocess_effectoro	Format raw EffectorO output
preprocess_deepredeff	Format raw deepredeff output
preprocess_wideeffhunter	Format WideEffHunter predictions from FASTA files
preprocess_t3sepp	Format T3SEpp output
merge_predictions	Merge formatted prediction results into one CSV
encode_predictions	Encode predictions into model input format
predict	Run the trained deep learning model on encoded input

Required options

• input - Input file path
• output - Output file path
• --kingdom - fungi, oomycete or bacteria
• --pred - predicted effector output from WideEffHunter (only used for WideEffHunter)

Special cases

• WideEffHunter: Requires the complete input files and predicted output file from WideEffHunter program.
• Merge: Accepts multiple input files followed by output file

1. Preprocess individual tool outputs

Process raw outputs from various effector prediction programs:

# Format EffectorP-3.0 results
fimep preprocess_effectorp --input effectorp_result.txt --output formatted_ep_output.csv --kingdom fungi


# Format deepredeff results
fimep preprocess_deepredeff --input deepredeff_result.csv --output formatted_dr_output.csv --kingdom fungi


# Format WideEffHunter results 
fimep preprocess_wideeffhunter --input complete_seq_file.fasta --pred predicted_wideeffhunter_output.fasta --output formatted_we_output.csv --kingdom oomycete


# Format T3SEpp results (bacteria only) 
fimep preprocess_t3sepp --input t3sepp_result.txt --output formatted_t3p_output.csv --kingdom bacteria


# Format EffectiveT3 results (bacteria only)
fimep preprocess_effectivet3 --input effectiveT3_result.csv --output formatted_et3_output.csv --kingdom bacteria


# Format EffectorO results (oomycete only)
fimep preprocess_effectoro --input effectoro_result.csv --output formatted_eo_output.csv --kingdom oomycete

2. Merge formatted predictions

Combine multiple prediction files for fungal pathogens into a single dataset:

fimep merge_prediction --input formatted_dr_output.csv formatted_et3_output.csv formatted_t3p_output.csv --output merged_data.csv

3. Encode merged data

Encode and scale the merged predictions for model input:

fimep encode --input merged_data.csv --output encoded_output.csv --kingdom fungi

4. Generate final prediction

fimep predict --input encoded_input.csv --output final_predictions.csv

---

Output format

The final output is a CSV file with two columns:

• Identifier: Sequence identifier
• Pred_Label: Final prediction (Effector/Non-Effector)

Example:

Identifier	Pred_Label
seq1	Effector
seq2	Non-Effector
seq3	Effector

Documenation

1. Extract predicted effectors as a fasta file for other downstream analysis

---

Contact

For issues, questions or contributions, please open an issue on the GitHub repository or contact us via email.

---

License

MIT License © 2025 Love Odunlami