FeatureDCAData

This repository contains supporting data for the FeatureDCA project, organized into three main folders. Each folder contains data relevant to the evolutionary analysis and mutational scanning of protein families.

---

Folder Structure

data/

Contains multiple protein family datasets, originally sourced from the ArDCAData repository, used as reference for model training and comparison:

• FASTA files and related sequence data.
• Experimental PDB files that can be used for structural comparison

RR_homodimers/

Data specific for the analysis of the Response Regulator (RR) protein family.

• *.hmm : Hidden Markov Models used for sequence alignment.
• *.fasta: Original and aligned sequences relative to classes Trans_Reg_C, GerE, and LytTR of the RR family .
• generated_sequences/: This folder contains sequences generated by FeatureDCA using varying numbers of principal components (d), conditioned to the three prototypical sequence classes.
• experimental_PDBs/: This folder contains the experimental PDBs relative to the three classes: 1NXS(Trans_Reg_C, 4ZMS(GerE), and 4CBV(LytTR).
• predicted_PDBs/: This folder contains the results of AlphaFold3 relative the generated sequences.

DMS_beta_lactamase/

Contains data related to Beta-lactamase deep mutational scanning (DMS), originally curated by the Ostermeier lab.

• *.fasta: Wild-type and variant sequences.
• *.hmm : Hidden Markov Models used for sequence alignment.
• *.xlsx : Preprocessed DMS dataset with mutational effects and associated metadata.