Find open-source science resources

This package provides functions to standardise the analysis of Differential Allelic Representation (DAR). DAR compromises the integrity of Differential Expression analysis results as it can bias expression, influencing the classification of genes (or transcripts) as being differentially expressed. DAR analysis results in an easy-to-interpret value between 0 and 1 for each genetic feature of interest, where 0 represents identical allelic representation and 1 represents complete diversity. This metric can be used to identify features prone to false-positive calls in Differential Expression analysis, and can be leveraged with statistical methods to alleviate the impact of such artefacts on RNA-seq data.

tidySpatialExperiment provides a bridge between the SpatialExperiment package and the tidyverse ecosystem. It creates an invisible layer that allows you to interact with a SpatialExperiment object as if it were a tibble; enabling the use of functions from dplyr, tidyr, ggplot2 and plotly. But, underneath, your data remains a SpatialExperiment object.

Open-ended AI scientist for computational fluid dynamics that spans literature-grounded ideation, OpenFOAM execution via Foam-Agent, vision-language physics verification of rendered flow fields, source-code modification for new physical models, and figure-grounded LaTeX manuscript writing within a single inspectable workflow (43+ stars, Python)

To render LaTex in your README, wrap the code in \\( and \\). Example: \\(\frac{1}{2}\\)

A package for accessing data from the NIST webbook...

This model is a domain-specific large language model fine-tuned from QwQ-32B, specialized in the field of perovskite solar cells, particularly focusing on precursor additives. It is designed to assist researchers, engineers, and material scientists by providing knowledge, insights, and suggestions…

A representation of variables appearing in models in the environmental research space.

SpatialArtifacts provides a data-driven two-step workflow to identify, classify, and handle spatial artifacts in spatial transcriptomics data. The package combines median absolute deviation (MAD)-based outlier detection with morphological image processing (fill, outline, and star patterns) to detect edge and interior artifacts. It supports multiple platforms including 10x Genomics Visium (standard and HD), allowing for consistent quality control across different spatial resolutions.

Target-Conditioned Molecular Ideation Model for Drug Discovery Research

Splatter is a package for the simulation of single-cell RNA sequencing count data. It provides a simple interface for creating complex simulations that are reproducible and well-documented. Parameters can be estimated from real data and functions are provided for comparing real and simulated datasets.

Genos-m is a foundation model for human-associated microbial genomes. It is trained to model microbial DNA sequences at single-nucleotide resolution and supports ultra-long genomic contexts up to one million tokens.

This package reads remote parquet files that have processed Bioconductor build report logs. Users may query the tables directly for specific information or use pre-defined helper functions for common queries. The logs processed are from https://bioconductor.org/checkResults/. In the future we will extend this package out to include processing of r-universe logs.

iSEEtree is an extension of iSEE for the TreeSummarizedExperiment data container. It provides interactive panel designs to explore hierarchical datasets, such as the microbiome and cell lines.

- Molecular Manipulation Made Easy. A light wrapper build on top of RDKit.

TRIP is a software framework that provides analytics services on antigen receptor (B cell receptor immunoglobulin, BcR IG | T cell receptor, TR) gene sequence data. It is a web application written in R Shiny. It takes as input the output files of the IMGT/HighV-Quest tool. Users can select to analyze the data from each of the input samples separately, or the combined data files from all samples and visualize the results accordingly.

StatescopeR is an R wrapper around Statescope, a computational framework designed to discover cell states from cell type-specific gene expression profiles inferred from bulk RNA profiles.

Implements miscellaneous functions for interpretation of single-cell RNA-seq data. Methods are provided for assignment of cell cycle phase, detection of highly variable and significantly correlated genes, identification of marker genes, and other common tasks in routine single-cell analysis workflows.

Convert AMBER forcefields from ANTECHAMBER to GROMACS format.

Unified pre-trained model for general physics simulation via lifted geometric pre-training, augmenting static geometry with synthetic dynamics to enable dynamics-aware self-supervision without physics labels; improves industrial-fidelity benchmarks spanning fluid mechanics and solid mechanics while reducing labeled data requirements by 20–60% (Physics-Scaling, 224+ stars)

Minimap2 is an pairwise aligner for genomic and spliced nucleotide sequences. It can perform the assembly-to-assembly alignment, and works with gzip'd FASTQ, FASTA formats. It also finds overlaps between long-reads.

The Chromosome Ontology is an automatically derived ontology of chromosomes and chromosome parts.

Data repository for SynPlanner — an open-source tool for retrosynthetic planning.

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]

DeeDeeExperiment is an S4 class extending the SingleCellExperiment class, designed to integrate and manage omics analysis results. It introduces two dedicated slots to store Differential Expression Analysis (DEA) results and Functional Enrichment Analysis (FEA) results, providing a structured approach for downstream analysis.

Fit a latent embedding multivariate regression (LEMUR) model to multi-condition single-cell data. The model provides a parametric description of single-cell data measured with treatment vs. control or more complex experimental designs. The parametric model is used to (1) align conditions, (2) predict log fold changes between conditions for all cells, and (3) identify cell neighborhoods with consistent log fold changes. For those neighborhoods, a pseudobulked differential expression test is conducted to assess which genes are significantly changed.

miaDash provides a Graphical User Interface for the exploration of microbiome data. This way, no knowledge of programming is required to perform analyses. Datasets can be imported, manipulated, analysed and visualised with a user-friendly interface.

Implements R bindings to C++ code for analyzing single-cell (expression) data, mostly from various libscran libraries. Each function performs an individual step in the single-cell analysis workflow, ranging from quality control to clustering and marker detection. Additional wrappers are provided for easy construction of end-to-end workflows involving Bioconductor objects like SingleCellExperiments.

Knowledge graph-guided synthetic data generation for LLM fine-tuning, achieving strong performance on scientific QA (GPQA-Diamond) and math reasoning (AIME)

This R package supports the handling and analysis of imaging mass cytometry and other highly multiplexed imaging data. The main functionality includes reading in single-cell data after image segmentation and measurement, data formatting to perform channel spillover correction and a number of spatial analysis approaches. First, cell-cell interactions are detected via spatial graph construction; these graphs can be visualized with cells representing nodes and interactions representing edges. Furthermore, per cell, its direct neighbours are summarized to allow spatial clustering. Per image/grouping level, interactions between types of cells are counted, averaged and compared against random permutations. In that way, types of cells that interact more (attraction) or less (avoidance) frequently than expected by chance are detected.

Qwen3-8B-syco_med-gated-attention-FT is a plug-and-play gated attention weight released for AI safety research.

All-atom biomolecular structure prediction for protein-nucleic acid-small molecule-metal ion complexes, enabling accurate modeling of covalent modifications and assemblies beyond proteins (Baker Lab, Science 2024)

The package allows users to readily import spatial data obtained from either the 10X website or from the Space Ranger pipeline. Supported formats include tar.gz, h5, and mtx files. Multiple files can be imported at once with *List type of functions. The package represents data mainly as SpatialExperiment objects.

Suite of tools to handle gene annotations in any GTF/GFF format.

Provides an R interface for various subsampling algorithms implemented in python packages. Currently, interfaces to the geosketch and scSampler python packages are implemented. In addition it also provides diagnostic plots to evaluate the subsampling.

This package allows to estimate chronological and gestational DNA methylation (DNAm) age as well as biological age using different methylation clocks. Chronological DNAm age (in years) : Horvath's clock, Hannum's clock, BNN, Horvath's skin+blood clock, PedBE clock and Wu's clock. Gestational DNAm age : Knight's clock, Bohlin's clock, Mayne's clock and Lee's clocks. Biological DNAm clocks : Levine's clock and Telomere Length's clock.

An ontology demonstrating rich ontology for rubber extrusion.

LoRA fine-tune of Qwen3.5-9B on synthetic clinical triage Q&A pairs generated from PubMed Central open-access papers. The model is specialized for emergency-medicine decision-making: triaging patients, applying clinical decision rules, and generating protocol-grounded triage recommendations.

Base model: google/gemma-4-26b-it Architecture: MoE — 26B total / ≈4B active parameters (1 shared expert + 8 routed from a pool of 128 per MoE layer, 30 MoE layers) Method: Activation-directed expert surgery — 128 → 64 experts per layer (50% reduction) Quantization: Q4KM (≈9.7 GB on disk) Tags:…

Base model: google/gemma-4-26b-it Architecture: MoE — 26B total / ≈4B active parameters (1 shared expert + 8 routed from a pool of 128 per MoE layer, 30 MoE layers) Method: Activation-directed expert surgery — 128 → 64 experts per layer (50% reduction) Quantization: Q4KM (≈9.7 GB on disk) Tags:…

Base model: google/gemma-4-26b-it Architecture: MoE — 26B total / ≈4B active parameters (1 shared expert + 8 routed from a pool of 128 per MoE layer, 30 MoE layers) Method: Activation-directed expert surgery — 128 → 64 experts per layer (50% reduction) Quantization: Q4KM (≈9.7 GB on disk) Tags:…

A small <720Kb C++ windows utility. That allows you to load Ancestry, 23andMe, FTDNA, or Genes for Good RAW DNA files search them, merge them. covert them to Ancestry format. But also create files from peer reviewed publications to compare with you loaded data to give your genetic disposition for the condition you have entered the data for an statistical risk if OR values are included. Included with the program are example files for Type 2 Diabetes risk factors. (As I have type 2 Diabetes so I could test the results).

Parallel symbolic regression network evaluating millions of expressions on GPU with automated subtree reuse, Nature Computational Science cover article (MIT, 2026)

Useful functions to visualize single cell and spatial data. It supports visualizing 'Seurat', 'SingleCellExperiment' and 'SpatialExperiment' objects through grammar of graphics syntax implemented in 'ggplot2'.

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

SEMPLR computes transcription factor binding affinity scores for genomic positions and genetic variants. Scores are computed from SNP Effect Matrices (SEMs) produced by SEMpl. 223 pre-computed SEMs are included with the package or custom sets can be provided. Enrichment can be tested among sets of genomic positions to determine if transcription factor binding events occur more often than expected. Comparing binding affinity scores between alleles can reveal differences in transcription factor binding with genetic variation. This package also includes several visualization functions to view scores both on the motif and variant/position level.

Deployable biomedical deep-research agent blueprint combining on-prem multimodal RAG, report generation, human-in-the-loop editing, and virtual screening with MolMIM and DiffDock for drug discovery workflows (2025)

- 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…

- 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…