Models#
Models can be viewed as sequence transforms or measures. They operate on a single entity—typically a biological sequence (or, in some cases, a 3D structure) - and map it to another representation or compute a value.
This distinguishes them from metrics: while a measure assigns a value to a single entity, a metric quantifies a relationship (e.g., distance or similarity) between two or more entities.
In practice, models transform biological sequences into feature representations or derive specific properties from them.
Overview#
The seqme library provides a suite of models that convert biological sequences into numerical representations. These models fall into two main categories:
Embedding models — map sequences to fixed-length vector representations.
Property models — compute scalar values or descriptors such as charge, weight, or hydrophobicity.
Additional third-party models are available at: szczurek-lab/seqme-thirdparty.
Models#
Wrapper for the ESM2 protein/peptide embedding model. |
|
ESMFold protein language model. |
|
A language model trained on RNA sequences, which computes sequence-level embeddings by averaging token embeddings. |
|
GENA-LM is a family of Open-Source Foundational Models for Long DNA Sequences trained on human DNA sequence. |
|
Computes normalized k-mer frequency embeddings for sequences. |
|
Aliphatic index of amino acid sequences. |
|
Aromaticity of amino acid sequences. |
|
Boman index, estimating binding potential to proteins. |
|
Net charge of amino acid sequences at a given pH. |
|
GRAVY (hydropathy) score for amino acid sequences. |
|
Hydrophobicity using a selected scale. |
|
Hydrophobic moment (i.e., amphiphilicity) for one or more amino acid sequences using a sliding-window approach. |
|
Instability index, predicting in vitro protein stability. |
|
Isoelectric point of amino acid sequences. |
|
Molecular weight of amino acid sequences. |
|
Lipophilicity for SMILES sequences. |
|
Quantitative Estimate of Drug-likeness for SMILES sequences. |
|
Synthetic Accessibility Score (SA Score) for SMILES sequences. |
|
Chemical validity for SMILES sequences. |
Miscellaneous#
These models perform special-purpose operations or combine multiple models.
Wrapper for loading and calling a third-party plugin model. |
|
Utility projecting sequence embeddings onto principle components. |
|
Combines multiple predictor functions into a weighted ensemble. |
Supported sequence types#
At-a-glance matrix of all models and supported sequence types.
— supported, 
— not supported
Model |
Protein |
Peptide |
RNA |
DNA |
Small Molecule |
|---|---|---|---|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|