Rapid and low-cost parsed variant library generation
uSort-M converts pooled DNA libraries into large collections of individually-isolated, sequence-verified variants at a fraction of traditional gene synthesis costs.
Traditional approaches to generating parsed variant libraries require expensive per-gene synthesis and individual cloning. uSort-M uses FACS to isolate single cells from a pooled transformation, then identifies variants by amplicon sequencing with well-specific barcodes.
Key advantages:
- Significant cost savings compared to traditional gene synthesis
- 10-day working time from oligo pool to verified clones
- Scalable from tens to thousands of variants
- Compatible with diverse library inputs
# Basic installation
pip install -e .
# Full installation with all dependencies
pip install -e ".[all]"The usortm demux command requires these tools to be installed separately:
| Tool | Min Version | Purpose | Install |
|---|---|---|---|
| dorado | 1.3+ | Barcode demultiplexing | GitHub releases |
| minimap2 | 2.20+ | Reference alignment | brew install minimap2 or conda install minimap2 |
| samtools | 1.16+ | BAM processing & consensus | brew install samtools or conda install samtools |
usortm auto-discovers dorado in common locations (~/Downloads/dorado-*/bin/, ~/.dorado/bin/). You can also set DORADO_PATH, MINIMAP2_PATH, or SAMTOOLS_PATH environment variables.
usortm estimate --library-size 500 --seq-length 300# 1. Initialize project from variant list
usortm plan variants.csv --output my_project/
# 2. [Perform wet lab: assembly, sorting, barcoding, sequencing]
# 3. Process sequencing data (with library CSV for variant calling)
usortm demux my_project/ --fastq sequencing-data.fastq --library-csv variants.csv
# 4. Generate hit-picking list
usortm pick my_project/
# 5. Create final report
usortm report my_project/| Command | Description |
|---|---|
estimate |
Quick cost and effort estimation |
plan |
Initialize project from variant list |
demux |
Demultiplex sequencing data (LevSeq barcodes via dorado, reference alignment, consensus, variant calling) |
pick |
Generate Integra ASSIST hit-picking list (ordered by input library) |
report |
Generate final plate maps, coverage stats, and HTML summary |
integra |
Standalone hit-list generation (without project) |
usortm estimate -n 500 -l 300
# Output:
# ╭────────────────────────────────╮
# │ uSort-M Cost Estimate │
# │ Library: 500 variants × 300 bp │
# ╰────────────────────────────────╯
#
# Cost Breakdown
# ╭────────────────────────┬─────────┬─────────────╮
# │ Step │ uSort-M │ Traditional │
# ├────────────────────────┼─────────┼─────────────┤
# │ Synthesis │ $1,373 │ $17,500 │
# │ Cloning │ $54 │ $6,048 │
# │ Sorting │ $104 │ N/A │
# │ Barcoding + Sequencing │ $1,477 │ $500 │
# │ Hit-picking │ $80 │ N/A │
# │ Total │ $3,088 │ $24,048 │
# ╰────────────────────────┴─────────┴─────────────╯
#
# 7.8-fold savings with uSort-M| Day | Step | Duration |
|---|---|---|
| 1 | Pooled assembly + transformation | 4-6 hours |
| 2+ | FACS sorting | ~8 min/plate |
| 2+ | Outgrowth | Overnight |
| 3+ | PCR barcoding | ~50 min/plate |
| 4-6 | Sequencing | 1-3 days |
| 6+ | Analysis + hit-picking | 1-2 hours |
from usortm.costs import cost_functions as cf
# Calculate costs
costs = cf.usortm_total_cost(
library_sizes=[500, 1000],
seq_lengths=[300]
)
# Run coverage simulations
from usortm.simulate import sortm
results = sortm.sortm(
n_sims=1000,
lib_size=500,
skew=4,
fold_sampling=8,
)Full documentation available at fordycelab.github.io/usortm
If you use uSort-M in your research, please cite:
Olivas MB, Almhjell PJ, Shanahan JD, Fordyce PM. uSort-M: Scalable isolation
of user-defined sequences from diverse pooled libraries. bioRxiv (2026). DOI: 10.64898/2026.01.12.699065
MIT License - see LICENSE for details.