Find open-source science resources

Topological pathway analysis tool able to integrate multi-omics data. It finds survival-associated modules or significant modules for two-class analysis. This tool have two main methods: pathway tests and module tests. The latter method allows the user to dig inside the pathways itself.

Autonomous pipeline from literature review→hypothesis→algorithm implementation→publication-level writing with Scientist-Bench evaluation

This package compares genomic positions and genomic ranges from multiple experiments to extract common regions. The size of the analyzed region is adjustable as well as the number of experiences in which a feature must be present in a potential region to tag this region as a consensus region. In genomic analysis where feature identification generates a position value surrounded by a genomic range, such as ChIP-Seq peaks and nucleosome positions, the replication of an experiment may result in slight differences between predicted values. This package enables the conciliation of the results into consensus regions.

DermLIP is a vision-language model for dermatology, trained on the Derm1M dataset—the largest dermatological image-text corpus to date. This model variant (PanDerm-base-w-PubMed-256) utilizes domain-specific pretraining to deliver superior performance compared to other DermLIP variants..

Chromatin looping is an essential feature of eukaryotic genomes and can bring regulatory sequences, such as enhancers or transcription factor binding sites, in the close physical proximity of regulated target genes. Here, we provide sevenC, an R package that uses protein binding signals from ChIP-seq and sequence motif information to predict chromatin looping events. Cross-linking of proteins that bind close to loop anchors result in ChIP-seq signals at both anchor loci. These signals are used at CTCF motif pairs together with their distance and orientation to each other to predict whether they interact or not. The resulting chromatin loops might be used to associate enhancers or transcription factor binding sites (e.g., ChIP-seq peaks) to regulated target genes.

The `miaTime` package provides tools for microbiome time series analysis based on (Tree)SummarizedExperiment infrastructure.

This package aims to import, parse, and analyze KEGG data such as KEGG PATHWAY and KEGG MODULE. The package supports visualizing KEGG information using ggplot2 and ggraph through using the grammar of graphics. The package enables the direct visualization of the results from various omics analysis packages.

Performs both stepwise and backward heuristic search for candidate (epi)genetic drivers based on a binary multi-omics dataset. CaDrA's main objective is to identify features which, together, are significantly skewed or enriched pertaining to a given vector of continuous scores (e.g. sample-specific scores representing a phenotypic readout of interest, such as protein expression, pathway activity, etc.), based on the union occurence (i.e. logical OR) of the events.

Hulu-Med: A Transparent Generalist Model towards Holistic Medical Vision-Language Understanding

# Biomni-R0-32B-Preview This repo contains the weights of Biomni-R0-32B-Preview, a research preview of the series of biomedical AI agents trained by the Biomni team.

The CTexploreR package re-defines the list of Cancer Testis/Germline (CT) genes. It is based on publicly available RNAseq databases (GTEx, CCLE and TCGA) and summarises CT genes' main characteristics. Several visualisation functions allow to explore their expression in different types of tissues and cancer cells, or to inspect the methylation status of their promoters in normal tissues.

Spatial-eXpression-R (spacexr) is a package for analyzing cell types in spatial transcriptomics data. This implementation is a fork of the spacexr GitHub repo (https://github.com/dmcable/spacexr), adapted to work with Bioconductor objects. The original package implements two statistical methods: RCTD for learning cell types and CSIDE for inferring cell type-specific differential expression. Currently, this fork only implements RCTD, which learns cell type profiles from annotated RNA sequencing (RNA-seq) reference data and uses these profiles to identify cell types in spatial transcriptomic pixels while accounting for platform-specific effects. Future releases will include an implementation of CSIDE.

The signeR package provides an empirical Bayesian approach to mutational signature discovery. It is designed to analyze single nucleotide variation (SNV) counts in cancer genomes, but can also be applied to other features as well. Functionalities to characterize signatures or genome samples according to exposure patterns are also provided.

InstaNovoPlus is a diffusion-based model for de novo peptide sequencing from mass spectrometry data. This model leverages multinomial diffusion for accurate, database-free peptide identification for large-scale proteomics experiments.

Support harvesting of diverse bioinformatic ontologies, making particular use of the ontologyIndex package on CRAN. We provide snapshots of key ontologies for terms about cells, cell lines, chemical compounds, and anatomy, to help analyze genome-scale experiments, particularly cell x compound screens. Another purpose is to strengthen development of compelling use cases for richer interfaces to emerging ontologies.

This package simulates regulations of ceRNA (Competing Endogenous) expression levels after a expression level change in one or more miRNA/mRNAs. The methodolgy adopted by the package has potential to incorparate any ceRNA (circRNA, lincRNA, etc.) into miRNA:target interaction network. The package basically distributes miRNA expression over available ceRNAs where each ceRNA attracks miRNAs proportional to its amount. But, the package can utilize multiple parameters that modify miRNA effect on its target (seed type, binding energy, binding location, etc.). The functions handle the given dataset as graph object and the processes progress via edge and node variables.

The aim of vidger is to rapidly generate information-rich visualizations for the interpretation of differential gene expression results from three widely-used tools: Cuffdiff, DESeq2, and edgeR.

The aim of TCGAbiolinks is : i) facilitate the GDC open-access data retrieval, ii) prepare the data using the appropriate pre-processing strategies, iii) provide the means to carry out different standard analyses and iv) to easily reproduce earlier research results. In more detail, the package provides multiple methods for analysis (e.g., differential expression analysis, identifying differentially methylated regions) and methods for visualization (e.g., survival plots, volcano plots, starburst plots) in order to easily develop complete analysis pipelines.

Seamlessly interfaces the Basic Local Alignment Search Tool (BLAST) running locally to search genetic sequence data bases. This work was partially supported by grant no. R21HG005912 from the National Human Genome Research Institute.

Typhoon-Si-Med-Thinking-4B is Southeast Asia’s first state-of-the-art, small, and efficient medical reasoning model, jointly developed by Typhoon (SCB 10X) and the Siriraj Informatics and Data Innovation Center (SiData+) at Siriraj Hospital, Mahidol University.

State-specific protein-ligand complex structure prediction with a multi-scale deep generative model, enabling conformational state-aware modeling of molecular interactions (329+ stars, 2024)

Multi-omic Pathway Analysis of Cells (MPAC), integrates multi-omic data for understanding cellular mechanisms. It predicts novel patient groups with distinct pathway profiles as well as identifying key pathway proteins with potential clinical associations. From CNA and RNA-seq data, it determines genes’ DNA and RNA states (i.e., repressed, normal, or activated), which serve as the input for PARADIGM to calculate Inferred Pathway Levels (IPLs). It also permutes DNA and RNA states to create a background distribution to filter IPLs as a way to remove events observed by chance. It provides multiple methods for downstream analysis and visualization.

EpiTxDb facilitates the storage of epitranscriptomic information. More specifically, it can keep track of modification identity, position, the enzyme for introducing it on the RNA, a specifier which determines the position on the RNA to be modified and the literature references each modification is associated with.

The package imports the result of tRNAscan-SE as a GRanges object.

The Mobilome Ontology describes bacterial mobile genetic elements, their mechanisms, and associated epidemiology.

Generalist foundation model and database for open-world medical image segmentation, enabling universal segmentation of diverse anatomical structures and pathologies with zero-shot generalization to unseen tasks and modalities (Nature Biomedical Engineering 2025)

Microbiome time series simulation with generalized Lotka-Volterra model, Self-Organized Instability (SOI), and other models. Hubbell's Neutral model is used to determine the abundance matrix. The resulting abundance matrix is applied to (Tree)SummarizedExperiment objects.

MedVAL-4B (medical text validator) is a language model fine-tuned to assess AI-generated medical text outputs at near physician-level reliability.

Enables the interactive visualization of dimensional reduction, clustering, and cell properties for scRNA-Seq results. It generates an interactive HTML page using either a numeric matrix, SummarizedExperiment, SingleCellExperiment or Seurat objects as input. The input data can be projected into two-dimensional representations by applying dimensionality reduction methods such as PCA, MDS, t-SNE, UMAP, and NMF. Displaying multiple dimensionality reduction results within the same interface, with interconnected graphs, provides different perspectives that facilitate accurate cell classification. The package also integrates unsupervised clustering techniques, whose results that can be viewed interactively in the graphical interface. In addition to visualization, this interface allows manual selection of groups, labeling of cell entities based on processed meta-information, generation of new graphs displaying gene expression values for each cell, sample identification, and visual comparison of samples and clusters.

LLM agent system synthesizing Wikipedia-like long-form research articles from scratch through multi-perspective question asking, web retrieval, and citation-grounded report generation, with Co-STORM extension for collaborative human-LLM knowledge curation conversations (Stanford OVAL, NAACL 2024 & EMNLP 2024)

This package aims to integrate GWAS-derived SNPs and coexpression networks to mine candidate genes associated with a particular phenotype. For that, users must define a set of guide genes, which are known genes involved in the studied phenotype. Additionally, the mined candidates can be given a score that favor candidates that are hubs and/or transcription factors. The scores can then be used to rank and select the top n most promising genes for downstream experiments.

AI-assisted mutation nomination approach optimizing protein function by integrating structural and evolutionary constraints into protein inverse folding models, compatible with ProteinMPNN, LigandMPNN, ESM-IF1, and SaProt (Chinese Academy of Sciences, 359+ stars)

We present a statistical simulator, scDesign3, to generate realistic single-cell and spatial omics data, including various cell states, experimental designs, and feature modalities, by learning interpretable parameters from real data. Using a unified probabilistic model for single-cell and spatial omics data, scDesign3 infers biologically meaningful parameters; assesses the goodness-of-fit of inferred cell clusters, trajectories, and spatial locations; and generates in silico negative and positive controls for benchmarking computational tools.

BANDITS is a Bayesian hierarchical model for detecting differential splicing of genes and transcripts, via differential transcript usage (DTU), between two or more conditions. The method uses a Bayesian hierarchical framework, which allows for sample specific proportions in a Dirichlet-Multinomial model, and samples the allocation of fragments to the transcripts. Parameters are inferred via Markov chain Monte Carlo (MCMC) techniques and a DTU test is performed via a multivariate Wald test on the posterior densities for the average relative abundance of transcripts.

Automated and rigorous experiments using AI agents for scientific discovery

This model is a lightweight model pre-trained on SELFIES (Self-Referencing Embedded Strings) representations of molecules. It is trained on 2.7M unique and valid molecules taken from COCONUTDB and ChemBL34, with 7.3M total generated masked examples.

First open-source AI-generated gene editing systems developed with protein language models, enabling programmable CRISPR-Cas nucleases for synthetic biology and therapeutic genome editing (Profluent, 2024)

> [!NOTE] > This model has been optimized using NVIDIA's TransformerEngine > library. Slight numerical differences may be observed between the original model and the optimized > model. For instructions on how to install TransformerEngine, please refer to the > official documentation.

> [!NOTE] > This model has been optimized using NVIDIA's TransformerEngine > library. Slight numerical differences may be observed between the original model and the optimized > model. For instructions on how to install TransformerEngine, please refer to the > official documentation.

Identify Surface Protein coding genes from a list of candidates. Systematically download data from GEO and TCGA or use your own data. Perform DGE on bulk RNAseq data. Perform Meta-analysis. Descriptive enrichment analysis and plots.

Partially latent flow matching model for the joint generation of a protein's amino acid sequence and full atomistic structure, including both backbone and side chains (2025)

RNAmodR provides classes and workflows for loading/aggregation data from high througput sequencing aimed at detecting post-transcriptional modifications through analysis of specific patterns. In addition, utilities are provided to validate and visualize the results. The RNAmodR package provides a core functionality from which specific analysis strategies can be easily implemented as a seperate package.

RNAmodR.RiboMethSeq implements the detection of 2'-O methylations on RNA from experimental data generated with the RiboMethSeq protocol. The package builds on the core functionality of the RNAmodR package to detect specific patterns of the modifications in high throughput sequencing data.

RNAmodR.AlkAnilineSeq implements the detection of m7G, m3C and D modifications on RNA from experimental data generated with the AlkAnilineSeq protocol. The package builds on the core functionality of the RNAmodR package to detect specific patterns of the modifications in high throughput sequencing data.

Tandem repeat genotyping with long reads, being a modified version of HipSTR.

Permutation analysis, based on Monte Carlo sampling, for testing the hypothesis that the number of conserved differentially methylated elements, between several generations, is associated to an effect inherited from a treatment and that stochastic effect can be dismissed.

Another list focuses on Python stuff related to Chemistry.