> For the complete documentation index, see [llms.txt](https://rexis.gitbook.io/rexis/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://rexis.gitbook.io/rexis/biological-sequences.md).

# Biological Sequences

### Biological Sequences

Biological sequences are a fundamental form of 1D biomedical data. They represent the linear order of molecular building blocks—such as nucleotides in DNA/RNA or amino acids in proteins.\
Analyzing these sequences is essential for understanding the structure, function, and evolution of living systems.

![](https://lh7-rt.googleusercontent.com/slidesz/AGV_vUfqLWj0NlNFmehEai_VLhtixxC9cSaB-3QFfjmpIercyFch7k4hnk0UQHAnP7zLq3c6ec8d9vNN07Ve63ypwMgHjSL-mcace7t7ca5oDgdys8v4OPZrATAfH0uuVbZYlpF9XJHnZw=s2048?key=_t06K9zDHrDgeHDIPqC4uaCR)

***

#### Genomic Sequences (DNA, RNA)

Genomic sequences encode the genetic blueprint of an organism. DNA (deoxyribonucleic acid) is the long-term store of genetic information, while RNA (ribonucleic acid) plays critical roles in gene expression and regulation.

DNA sequences use the bases: **A**, **T**, **C**, and **G**, while RNA replaces thymine (T) with **U** (uracil). Data formats commonly used include:

* **FASTA / FASTQ** – for raw or annotated sequences
* **SAM / BAM / CRAM** – for aligned reads
* **GFF / GTF** – for gene annotations

**Data Support**

Genomic data is large, complex, and sensitive.\
Our platform provides **a secure and scalable infrastructure** for storing, sharing, and analyzing such datasets. Using **Equivariant Encryption (EE)**, we enable:

* Encrypted genomic data repositories
* Collaborative analysis workflows
* Compliance with privacy and bioethics standards

**Model Support**

Foundation models have dramatically improved genomic analysis by enabling:

* Gene function prediction
* Regulatory element detection
* Variant impact modeling

We support **encrypted evaluation and fine-tuning of genomics foundation models**.

* [**scGPT**](https://github.com/bowang-lab/scGPT)\
  A generative pre-trained model for single-cell biology. It handles imputation, denoising, and synthetic state generation for gene expression data.
* [**DNABERT**](https://github.com/MAGICS-LAB/DNABERT_2)\
  A BERT-style model trained on human and bacterial genomes. It excels at tasks like promoter detection, splice site prediction, and regulatory element discovery.

***

#### Proteomic Sequences

Proteins are chains of amino acids that fold into 3D structures to carry out cellular functions.\
There are 20 standard amino acids, and their linear order determines a protein's folding and function.

Proteomic sequences are typically stored in:

* **FASTA** – for raw sequences
* **mzML / mzXML / mzIdentML** – for mass spectrometry data

**Data Support**

Proteomic data often includes sensitive therapeutic and diagnostic information.\
Our platform supports **secure storage and encrypted processing** of proteomic sequences, enabling:

* Secure data sharing between labs
* Privacy-preserving collaboration on biomarker discovery
* Scalable proteomics pipelines using EE

**Model Support**

Modern deep learning models are capable of:

* Predicting 3D protein structure
* Modeling protein interactions
* Designing novel therapeutics

We support **secure inference and fine-tuning** for proteomics foundation models.

* [**AlphaFold**](https://github.com/google-deepmind/alphafold)\
  A groundbreaking deep learning model that predicts the 3D structure of proteins from amino acid sequences with near-experimental accuracy. Its use cases span drug discovery, structural biology, and protein design.

***