KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh #dsh-plugin #ai-agents #developer-tools #typescript #codex #cli #claude-code #mcp #rust #deepseek-harness #ai-agent

devider

★ 46 Open GitHub ↗

Dividing heterogeneous long-read sequencing into groups with de Bruijn graphs

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# devider - long-read haplotypes from mixtures of "small" sequences

**devider** is a method that separates long reads (Nanopore or PacBio) into groups with similar alleles. This is called "phasing" or "haplotyping".

devider is a "local haplotyping" method, so it works best when the sequence-of-interest is approximately the size of the reads. For bacterial genome-scale haplotyping, consider another tool such as [floria](https://github.com/bluenote-1577/floria).

### Example use cases:

* mixed viral long-read samples (e.g. co-infections or quasispecies)
* amplicon/enriched sequencing of specific genes
* haplotyping small sections of multi-strain bacterial communities

<img width="600" height="200" src="https://github.com/user-attachments/assets/c0a82bb5-7feb-4d13-ab59-04da2bce52b3", caption="asdf">


 
High-depth, heterogeneous sequencing that spans a 1kb gene.

<img width="600" height="200" src="https://github.com/user-attachments/assets/34cb8bcf-8f23-47e4-b2f6-8515a21d3cf4", caption="asdf">


 
Separated groups ("haplotypes") after running devider.

### Why devider?

Similar tools exist for detection of similar haplotypes in mixtures. devider was developed to fill the following gaps:

* **Speed and low-memory** - devider scales approximately linearly with sequencing depth and # of SNPs. > 30,000x coverage genes can be haplotyped in minutes. 
* **High heterogeneity and coverage** - devider uses a de Bruijn Graph approach, which works with very diverse samples (> 10 haplotypes)
* **Ease-of-use + interpretable outputs** - conda installable, engineered in rust, simple command line. Outputs are easy to interpret (haplotagged BAM or MSA).

## Install

#### Conda (preferred)

```sh
mamba install -c bioconda devider
devider -h 
```

#### Static binary (only x86_64 architectures, without extra pipeline scripts)

```sh
wget https://github.com/bluenote-1577/devider/releases/download/latest/devider
chmod +x devider
./devider
```

See the [installation instructions on the wiki](https://github.com/bluenote-1577/devider/wiki/Installation) if you want want to compile devider (written in Rust) or you're **not** on x86-64 CPUs.

## Quick Start after install

### Option 1 (more flexible): Running devider with VCF + BAM
```sh
git clone https://github.com/bluenote-1577/devider
cd devider
devider -b hiv_test/3000_95_3.bam  -v hiv_test/3000_95_3.vcf.gz  -r hiv_test/OR483991.1.fasta -o devider_output

# results folder
ls devider_output
```
### Option 2 (easier): Running devider with reads

If installed from conda:
```sh
git clone https://github.com/bluenote-1577/devider
cd devider
run_devider_pipeline -i hiv_test/3000_95_3.fastq.gz -r hiv_test/OR483991.1.fasta -o devider_pipeline_output

# results folder
ls devider_pipeline_output

# intermediate files (bam + vcf files)
ls devider_pipeline_output/pipeline_files
```

If you **did not** install via conda and want to run the pipeline script, ensure the following are in `PATH`.

* tabix
* minimap2
* lofreq
* devider

Then run `scripts/run_devider_pipeline` in the GitHub repository.

## How to use devider

* [Cookbook](https://github.com/bluenote-1577/devider/wiki/Cookbook) - see here for usage examples.
* [Advanced usage manual](https://github.com/bluenote-1577/devider/wiki/Advanced-usage-manual) - see here for more detailed information about parameters and usage.
* [Output format](https://github.com/bluenote-1577/devider/wiki/Output-format) - for more information on how to interpret outputs.

## Citation

Long-read reconstruction of many diverse haplotypes with devider. Jim Shaw, Christina Boucher, Yun William Yu, Noelle Noyes, Heng Li. Genome Research (2025).

devider

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README