Salis Lab Promoter Calculator

The Salis Lab Promoter Calculator is a 346-parameter biophysical and machine-learning model from the Salis Lab that predicts the strength of σ70 housekeeping promoters in Escherichia coli. It scans both strands of an input DNA sequence for every candidate transcription start site, decomposes the surrounding region into the canonical promoter elements (UP element, −35 hexamer, spacer, −10 hexamer, and discriminator), and returns the predicted Gibbs free energy of RNA polymerase holoenzyme binding (dG_total) together with a calibrated transcription initiation rate (Tx_rate).

License: Salis Lab Promoter Calculator has a GPL-3.0 license. Please refer to the license for full terms.

This toolkit is open source. Any third-party models, product names, or trademarks referenced are the property of their respective owners, and Proto is not affiliated with them.

---

GitHub 6 GitHub 6 Publication Publication Cite Cite Tool Source Tool Source Open as Notebook Open as Notebook

barricklab/promoter-calculator

6 stars

View repo

Automated Model-Predictive Design of Synthetic Promoters to Control Transcriptional Profiles in Bacteria

Travis L. LaFleur, Ayaan Hossain and Howard M. Salis

Nature Communications (2022)

Read paper

@article{lafleur2022promoter,
  title={Automated Model-Predictive Design of Synthetic Promoters to Control Transcriptional Profiles in Bacteria},
  author={LaFleur, Travis L. and Hossain, Ayaan and Salis, Howard M.},
  journal={Nature Communications},
  volume={13},
  number={1},
  pages={5159},
  year={2022},
  doi={10.1038/s41467-022-32829-5}
}

Copy citation

proto-bio/proto-tools/proto_tools/tools/gene_annotation/promoter_calculator

View source

Open Notebook

Open notebook

Function	Description
run_promoter_calculator()	Predict E. coli sigma70 promoter strength (dG and Tx rate) on DNA sequences	Docs Source

Background

The Promoter Calculator (LaFleur et al., 2022) predicts site-specific σ70 transcription initiation rates from DNA sequence alone. σ70 is the housekeeping sigma factor that recruits E. coli RNA polymerase to the majority of constitutive promoters, and its strength is set by the sequence and geometry of the surrounding promoter elements. The model decomposes a candidate promoter into the UP element, the −35 hexamer, the spacer, the extended and core −10 hexamers, the discriminator, and the initial transcribed region, then fits 346 Ridge-regression coefficients that link these elements to an overall binding free energy ΔG_total. The fitted ΔG_total is mapped to an absolute transcription initiation rate via log(TX / TX_ref) = -β (ΔG_total - ΔG_total_ref), with the fit grounded in massively parallel reporter assay measurements. The released coefficients were trained on 5,193 designed promoters with a single dominant transcription start site and validated against 22,132 diverse bacterial σ70 promoters drawn from multiple datasets.

The reference Python implementation used here is the barricklab/promoter-calculator fork from the Barrick Lab, which packages the original Salis Lab algorithm for streamlined installation and adds optional multi-threading for the internal transcription-start-site scan.

Learning Resources

• barricklab/promoter-calculator (Barrick Lab) - the Python fork used here, with installation instructions and the command-line surface that the wrapper drives.
• hsalis/SalisLabCode (Promoter_Calculator) (Salis Lab) - the original reference implementation distributed alongside the publication and the source of the trained coefficients.
• salislab.net (Salis Lab) - the lab home page and entry point to the hosted Promoter Calculator web service.

Tools

Salis Lab Promoter Calculator (promoter-calculator)

Scans one or more DNA sequences on both strands for every candidate σ70 transcription start site and returns, per sequence, a list of PromoterPrediction rows. Each row carries the predicted TSS position and strand, the binding free energy dG_total (kcal/mol), the transcription initiation rate Tx_rate (arbitrary units), the DNA spanning the predicted promoter, and the start and end positions of the UP, −35, spacer, −10, and discriminator elements.

API Reference

Source

Input: PromoterCalculatorInput

sequences · List[string] · required

DNA sequences to scan for sigma70 promoters. Labeled positionally (seq_0, seq_1, …); results are returned in input order.

Source

Config: PromoterCalculatorConfig

threads · integer · default: "1"

CPU threads for the calculator’s internal parallelism. Default 1.

circular · boolean · default: "False"

Treat sequences as circular (examines the wraparound junction). Default False.

verbose · integer · default: "0"

Verbosity level (0=quiet, 1=info, 2=debug, 3=raw subprocess stderr). True is coerced to 1 and False to 0.

device · string · default: "cpu"

Device to run the tool on.

timeout · integer · default: "600"

Maximum execution time in seconds. None waits indefinitely.

seed · integer

Random seed. When set, tools run reproducibly up to small GPU float noise (see BaseToolOutput.approx_equal), and the seed participates in cache keys. When None, cacheable seed-sensitive tools skip cache until seeded.

Source

Output: PromoterCalculatorOutput

results · List[PromoterCalculatorSequenceResult]

Per-sequence predictions.

Show PromoterCalculatorSequenceResult

sequence_id · string · required

ID of the input sequence

predictions · List[PromoterPrediction]

Predicted promoters across both strands

Applications

Use this to quantify σ70 promoter strength when designing or analysing E. coli expression cassettes. Common workflows include ranking a synthetic library such as the Anderson collection of J231xx variants by predicted Tx_rate to pick a target strength, sweeping a candidate construct to flag unintended cryptic promoters that could drive off-target transcription before ordering DNA, and annotating native intergenic regions across an E. coli genome to estimate baseline σ70 activity. Pair the predicted transcription rate with downstream ribosome binding site strength and plasmid copy number when projecting end-to-end protein expression.

Usage Tips

• Set circular=True for plasmids and bacterial chromosomes. Linear scanning of a circular sequence cannot see candidates that span the wraparound origin. When the input is circular, the calculator examines the junction explicitly and recovers any promoter that straddles it.
• Short inputs need flanking context to score. The element scan needs roughly 20 nucleotides on either side of the promoter region; sequences shorter than the scan window return no predictions. Pad with neutral flanking sequence when scoring a single promoter element in isolation.
• Predictions are calibrated for E. coli σ70 only. Applying the model to other organisms or to alternative sigma factors (σS, σ32, σ54, σ28) is not validated. Treat any cross-organism output as a relative ranking, not as a calibrated rate.
• Tx_rate is transcription initiation, not protein expression. The model captures RNA polymerase binding and open-complex formation. End-to-end expression also depends on the ribosome binding site, mRNA stability, copy number, and growth state, none of which the calculator sees.
• The model does not see transcription factors, attenuation, or anti-σ factors. Repression by a TF bound near the promoter, riboswitch attenuation, and anti-σ sequestration all reduce in vivo output without changing the predicted Tx_rate. Use the prediction as the unregulated upper bound.

Toolkit Notes

These apply to every Promoter Calculator tool in this toolkit (promoter-calculator).

• Runs on CPU only. The model is a Ridge-regression sum over 346 coefficients that ship with the Python source. No neural network is loaded at inference time, and there is no GPU acceleration to enable.
• Self-contained after install. The standalone setup builds a Python virtual environment and pulls dependencies once; subsequent runs need no further network access and no model-weight downloads.
• threads parallelises the internal TSS scan within a single sequence. Raising it shortens wall-clock time on long inputs but does not change the predictions. Across input sequences the wrapper itself runs sequentially.

Example notebook: See the full working example for a copy-paste-ready walkthrough.

Infrastructure Guides

The following guides cover how to run tools efficiently and at scale.

Tool Persistence

Keep a tool’s model warm across calls instead of reloading it every invocation.

Device Management

How GPUs are allocated to tools and how to target specific devices.

Parallel Execution

Fan a batch of inputs out across multiple GPUs.

Cloud Inference

Run tools on managed cloud infrastructure with no local setup.