1. Description
2. Requirements on GFA file
3. Installation
4. Online Interaction (option)
5. Examples
6. Citations
7. Acknowledgements

GFA Kaleidos is a tool designed to analyze graph statistics using GFA format input. We characterized the GFA file by modeling it with 3 different graph representations: directed graph, bidirected graph, and biedged graph. For each model, we computed statistics on vertices, edges, and subgraphs.

The tool supports GFA format versions v1.0 and v1.1. Certain indicators may not function without path or walk information.

Currently, the software performs well on graphs generated by pggb and minigraph-cactus. If you encounter compatibility issues with graphs generated by other software, feel free to reach out to us.

The command-line version is now fully implemented. You can clone the repository, compile the code, and run it locally.

git clone git@github.com:YX-Xiang/GFAKaleidos.git
cd GFAKaleidos/V1
make -C ./algorithm clean
make -C ./algorithm -j
./algorithm/gfaKaleidos.exe INPUT_GFA -o OUTPUT_PATH -t THREAD_NUM
zip ZIP_NAME.zip OUTPUT_PATH

1. INPUT_GFA: The file path of the GFA file that the user wants to analyze.

2. OUTPUT_PATH: The output directory path where the results will be saved. This should be a folder.

3. THREAD_NUM: The number of threads to use for the analysis. The default value is 1.

4. ZIP_NAME: The file path for the compressed output result in .zip format (optional). If further interaction with the website is required, this parameter must be specified.

Upload ZIP_NAME.zip to web server, which offers a user-friendly interface with clear guidance. Visual outputs can be exported to HTML or PDF formats, facilitating sharing and dissemination.

Here below a guide on how to generate the datasets used for the experiments presented in the GFA Kaleidos paper, generate the Drosophila melanogaster pangenome using the 2 different tools and run the analysis. The final datasets 01_dm_pggb.gfa and 02_dm_mc.gfa can be found in the example folder.

To download the haplotypes, you can go to NCBI. It is recommended to use the ncbi-genome-download command to batch download genomic data, accession numbers are provided in example/accessions.txt

ncbi-genome-download --assembly-accessions example/accessions.txt invertebrate --formats fasta --output-folder example/genome -s genbank --species-taxid 7227 --flat-output -P

cd example/genome
gunzip *.fna.gz

In this step, we isolated chromosome 4 in the Drosophila assembly. You can run python ../scripts/separation.py directly.

# separation.py
import os

def process_fna(fna_path):
    with open(fna_path, 'r') as fna:
        sequences = fna.read().split('>')
        for sequence in sequences:
            if not sequence.strip():
                continue

            sequence_parts = sequence.split('\n', 1)
            header = sequence_parts[0].split(' ')
            chromosome = header[6]
            if chromosome == '4':
                sequence_data = ''.join(sequence_parts[1:]).strip()
                parts = os.path.basename(fna_path).split('_')
                new_file_name = parts[0] + '_' + parts[1] + '_' + chromosome + '.fa'
                new_file_path = os.path.join(os.path.dirname(fna_path), new_file_name)
                with open(new_file_path, 'w') as new_file:
                    new_file.write('>' + sequence_parts[0].strip() + '\n' + sequence_data)

current_directory = os.getcwd()
for file in os.listdir(current_directory):
    if file.endswith(".fna"):
        process_fna(os.path.join(current_directory, file))

After this, you will get 10 files with names like GCA_029775XXX_4.fa

First, ensure you have properly installed PGGB using conda with the environment name pggb.

Next, write the absolute path of the .fa file to pggb_link.txt in the following format:

genome/GCA_029775095.1_4.fa
genome/GCA_029775115.1_4.fa
...
genome/GCA_029775315.1_4.fa

Finally, execute the script pggb.sh.
Note that conda.sh must be soft-linked to the current script execution directory.

bash ../scripts/pggb.sh pggb_link.txt 10 01_dm_pggb 1 OUTPUT_FOLDER 

Download and install cactus. Make sure that there is BUILD_FOLDER/bin/activate.

Next, write the ID and absolute path of the .fa file to mc_link.txt in the following format:

GCA_029775095	genome/GCA_029775095.1_4.fa
GCA_029775115	genome/GCA_029775115.1_4.fa
...
GCA_029775315	genome/GCA_029775315.1_4.fa

Finally, execute the script minigraph_cactus.sh.

bash ../scripts/minigraph_cactus.sh mc_link.txt GCA_029775095 02_dm_mc 1 OUTPUT_FOLDER BUILD_FOLDER/bin/activate

Upload 01_dm_pggb.gfa and 02_dm_mc.gfa, and click all checkboxes in the window.

The final results are as follows:

Please cite our paper if you find GFAKaleidos useful:

Yixin XIANG, Keyu LIU, Leqi WANG, Jianyu ZHOU. GFAKaleidos: a tool for computing and comparing pangenome graph statistics. Front. Comput. Sci., 2026, 20(4): 2004903 DOI:10.1007/s11704-025-50324-0.

GFA Kaleidos uses many different algorithms and individual code contributions. We are particularly grateful to:

• Tarjan R. Depth-first search and linear graph algorithms. SIAM journal on computing, 1972, 1(2): 146–160
• Gabow H N. Path-based depth-rst search for strong and biconnected components. Information Processing Letters, 2000
• Szwarcfiter J L, Lauer P E. A search strategy for the elementary cycles of a directed graph. BIT Numerical Mathematics, 1976, 16(2): 192–204
• Parmigiani L, Garrison E, Stoye J, et al. Panacus: fast and exact pangenome growth and core size estimation. Bioinformatics, 2024, 40(12): btae720
• Dabbaghie F, Ebler J, Marschall T. BubbleGun: enumerating bubbles and superbubbles in genome graphs. Bioinformatics, 2022, 38(17): 4217-4219