Find open-source science resources

Perl package for circular plots, which are well suited for genomic rearrangements.

Our pipeline, MICSQTL, utilizes scRNA-seq reference and bulk transcriptomes to estimate cellular composition in the matched bulk proteomes. The expression of genes and proteins at either bulk level or cell type level can be integrated by Angle-based Joint and Individual Variation Explained (AJIVE) framework. Meanwhile, MICSQTL can perform cell-type-specic quantitative trait loci (QTL) mapping to proteins or transcripts based on the input of bulk expression data and the estimated cellular composition per molecule type, without the need for single cell sequencing. We use matched transcriptome-proteome from human brain frontal cortex tissue samples to demonstrate the input and output of our tool.

Educational resource on performing RNA-seq analysis in the cloud using Amazon AWS cloud services. Topics include preparing the data, preprocessing, differential expression, isoform discovery, data visualization, and interpretation.

FeatSeekR performs unsupervised feature selection using replicated measurements. It iteratively selects features with the highest reproducibility across replicates, after projecting out those dimensions from the data that are spanned by the previously selected features. The selected a set of features has a high replicate reproducibility and a high degree of uniqueness.

Programmatically access the NIH / NCI Genomic Data Commons RESTful service.

Multimodal deep learning framework integrating peptide-MHC protein sequence, structure, and biochemical properties to predict class-I immunogenicity for infectious disease epitopes and cancer neoepitopes with cancer-wildtype contrastive learning, enabling personalized vaccine design (Krishnaswamy Lab, Yale University)

R Package for interactive visualization and browsing NGS data. It contains a browser for both transcript and genomic coordinate view. In addition a QC and general metaplots are included, among others differential translation plots and gene expression plots. The package is still under development.

A client for BEDbase. bedbaser provides access to the API at api.bedbase.org. It also includes convenience functions to import BED files into GRanges objects and BEDsets into GRangesLists.

A toolkit for simulating differential microbiome data designed for longitudinal analyses. Several functional forms may be specified for the mean trend. Observations are drawn from a multivariate normal model. The objective of this package is to be able to simulate data in order to accurately compare different longitudinal methods for differential abundance.

This is the Provenance Information for Materials Science (PRIMA) Ontology, version 3.0, aligned with PMDco v3 and based on BFO (Basic Formal Ontology). This complete module imports all PRIMA modules (core, data-analysis-lifecycle, dataset, experiment, and computational) in their v3.0 versions. [from https://purls.helmholtz-metadaten.de/prima/complete]

R interface for importing and analyzing enzyme information from the BRENDA database.

Use hail via basilisk when appropriate, or via reticulate. This package can be used in terra.bio to interact with UK Biobank resources processed by hail.is.

ClonalSim generates realistic mutational profiles of tumor samples with hierarchical clonal structure. It simulates founder, shared, and private mutations with biologically realistic noise models including intra-tumor heterogeneity (Beta distribution) and technical sequencing noise (negative binomial depth variation, binomial read sampling, base errors). The package is designed for benchmarking variant callers, testing clonal deconvolution algorithms, and teaching tumor heterogeneity concepts.

Gemma 4 E2B fine-tuned on 225K drug–target pairs for novel small-molecule generation.

Composite-objective protein design framework integrating Boltz, AlphaFold2, OpenFold3, ProteinMPNN, and ESM via JAX-based gradient optimization over continuous relaxed sequence space for multi-property binder design (319+ stars, MIT License, 2025)

First system to make novel, verifiable scientific discoveries by pairing LLMs with evolutionary search, solving open problems in combinatorics (cap set problem) and discovering faster matrix multiplication algorithms

SCUDO (Signature-based Clustering for Diagnostic Purposes) is a rank-based method for the analysis of gene expression profiles for diagnostic and classification purposes. It is based on the identification of sample-specific gene signatures composed of the most up- and down-regulated genes for that sample. Starting from gene expression data, functions in this package identify sample-specific gene signatures and use them to build a graph of samples. In this graph samples are joined by edges if they have a similar expression profile, according to a pre-computed similarity matrix. The similarity between the expression profiles of two samples is computed using a method similar to GSEA. The graph of samples can then be used to perform community clustering or to perform supervised classification of samples in a testing set.

Automated and rigorous experiments using AI agents for scientific discovery

The Simplified Upper Level Ontology (SULO) is ontology with a minimal set of classes and relations to guide the development of a personal health knowledge graph. [from homepage]

- 2025-05-15: We identified a bug in the Bacformer Large code on HuggingFace which resulted in a significant drop in the quality of the output embeddings. This is now fixed, but if you downloaded or cached the model before this date, re-download and use the latest model revision before running…

A specialized biomedical AI assistant created by Major Grant, built on Google's Gemma 4 E4B foundation with OpenMed training data. GGUF format for efficient local inference.

An R package for integrated differential expression and differential network analysis based on omic data for cancer biomarker discovery. Both correlation and partial correlation can be used to generate differential network to aid the traditional differential expression analysis to identify changes between biomolecules on both their expression and pairwise association levels. A detailed description of the methodology has been published in Methods journal (PMID: 27592383). An interactive visualization feature allows for the exploration and selection of candidate biomarkers.

Highly multiplexed imaging acquires the single-cell expression of selected proteins in a spatially-resolved fashion. These measurements can be visualised across multiple length-scales. First, pixel-level intensities represent the spatial distributions of feature expression with highest resolution. Second, after segmentation, expression values or cell-level metadata (e.g. cell-type information) can be visualised on segmented cell areas. This package contains functions for the visualisation of multiplexed read-outs and cell-level information obtained by multiplexed imaging technologies. The main functions of this package allow 1. the visualisation of pixel-level information across multiple channels, 2. the display of cell-level information (expression and/or metadata) on segmentation masks and 3. gating and visualisation of single cells.

Use BridgeDb functions and load identifier mapping databases in R. It uses GitHub, Zenodo, and Figshare if you use this package to download identifier mappings files.

Machine learning model predicting cellular perturbation response across diverse contexts with State Transition (ST) and State Embedding (SE) variants, featuring CLI tooling, PyPI distribution, and Virtual Cell Challenge integration (575+ stars)

HOSO is an ontology of informational entities and processes related to healthcare organizations and services.

Biocaml aims to be a high-performance user-friendly library for Bioinformatics.

bayNorm is used for normalizing single-cell RNA-seq data.

This package defines interfaces from R to scvi-tools. A vignette works through the totalVI tutorial for analyzing CITE-seq data. Another vignette compares outputs of Chapter 12 of the OSCA book with analogous outputs based on totalVI quantifications. Future work will address other components of scvi-tools, with a focus on building understanding of probabilistic methods based on variational autoencoders.

The scDiagnostics package provides diagnostic plots to assess the quality of cell type assignments from single cell gene expression profiles. The implemented functionality allows to assess the reliability of cell type annotations, investigate gene expression patterns, and explore relationships between different cell types in query and reference datasets allowing users to detect potential misalignments between reference and query datasets. The package also provides visualization capabilities for diagnostics purposes.

Keras-based scientific neural networks

This package takes a list of p-values resulting from the simultaneous testing of many hypotheses and estimates their q-values and local FDR values. The q-value of a test measures the proportion of false positives incurred (called the false discovery rate) when that particular test is called significant. The local FDR measures the posterior probability the null hypothesis is true given the test's p-value. Various plots are automatically generated, allowing one to make sensible significance cut-offs. Several mathematical results have recently been shown on the conservative accuracy of the estimated q-values from this software. The software can be applied to problems in genomics, brain imaging, astrophysics, and data mining.

Medical time series foundation model pretrained on 454B time points from heterogeneous clinical corpora spanning ICU physiological signals and hospital EHR, with continuous-time rotary positional encoding, frequency-specialized Mixture-of-Experts, and neural ODE extrapolation for zero-shot forecasting across irregular and multimodal temporal health data (Microsoft, 399+ stars, MIT License)

Incremental knowledge graph construction using LLMs with entity extraction and Neo4j visualization

Translate differential transcript usage results into discrete splice events.

Abstract:

Scientific equation discovery and symbolic regression using LLMs, combining code generation with evolutionary search (ICLR 2025 Oral)

Directed message passing neural networks for property prediction of molecules and reactions with uncertainty and interpretation.

First architecture deeply integrating a DNA foundation model with an LLM for multimodal biological reasoning, achieving 98% accuracy on KEGG disease pathway prediction and 15%+ average gains on variant effect prediction with interpretable step-by-step reasoning traces (bowang-lab, 390+ stars)

Tools for manipulating paired ranges and working with Hi-C data in R. Functionality includes manipulating/merging paired regions, generating paired ranges, extracting/aggregating interactions from `.hic` files, and visualizing the results. Designed for compatibility with plotgardener for visualization.

This package aims to perform power analysis for the MeRIP-seq study. It calculates FDR, FDC, power, and precision under various study design parameters, including but not limited to sample size, sequencing depth, and testing method. It can also output results into .xlsx files or produce corresponding figures of choice.

Sparse Autoencoder (SAE) trained on residue-level embeddings from ESM-2 (650M, layer 33) for interpretability research on protein language models.

- 2025-05-15: We identified a bug in the Bacformer Large code on HuggingFace which resulted in a significant drop in the quality of the output embeddings. This is now fixed, but if you downloaded or cached the model before this date, re-download and use the latest model revision before running…

Evo 2 is a state-of-the-art DNA language model trained autoregressively on trillions of DNA tokens.

D3 JavaScript based genome viewer. Constructs SVGs.

GPU-accelerated differentiable physics simulation engine built on NVIDIA Warp, supporting rigid/soft body, cloth, and gradient-based optimization for scientific ML, initiated by Disney Research, DeepMind, and NVIDIA (Linux Foundation, Apache 2.0, 2025)

The Zarr specification defines a format for chunked, compressed, N-dimensional arrays. It's design allows efficient access to subsets of the stored array, and supports both local and cloud storage systems. Rarr aims to implement this specification in R with minimal reliance on an external tools or libraries.

This package provides an R wrapper for the popular Bowtie2 sequencing read aligner, optimized to run on NVIDIA graphics cards. It includes wrapper functions that enable both genome indexing and alignment to the generated indexes, ensuring high performance and ease of use within the R environment.