Find open-source science resources

Module for single-cell data extraction given a segmentation mask and multi-channel image.

The cfTools R package provides methods for cell-free DNA (cfDNA) methylation data analysis to facilitate cfDNA-based studies. Given the methylation sequencing data of a cfDNA sample, for each cancer marker or tissue marker, we deconvolve the tumor-derived or tissue-specific reads from all reads falling in the marker region. Our read-based deconvolution algorithm exploits the pervasiveness of DNA methylation for signal enhancement, therefore can sensitively identify a trace amount of tumor-specific or tissue-specific cfDNA in plasma. cfTools provides functions for (1) cancer detection: sensitively detect tumor-derived cfDNA and estimate the tumor-derived cfDNA fraction (tumor burden); (2) tissue deconvolution: infer the tissue type composition and the cfDNA fraction of multiple tissue types for a plasma cfDNA sample. These functions can serve as foundations for more advanced cfDNA-based studies, including cancer diagnosis and disease monitoring.

RAiSD-AI is a tool for training, testing, and deploying Convolutional Neural Networks to detect selective sweeps in genomic data, extending the functionality of the original RAiSD software with machine learning capabilities. It supports SNP data processing, CNN model training with TensorFlow or PyTorch, and genome-wide selective sweep detection.

Banksy is an R package that incorporates spatial information to cluster cells in a feature space (e.g. gene expression). To incorporate spatial information, BANKSY computes the mean neighborhood expression and azimuthal Gabor filters that capture gene expression gradients. These features are combined with the cell's own expression to embed cells in a neighbor-augmented product space which can then be clustered, allowing for accurate and spatially-aware cell typing and tissue domain segmentation.

The GenomicInteractionNodes package can import interactions from bedpe file and define the interaction nodes, the genomic interaction sites with multiple interaction loops. The interaction nodes is a binding platform regulates one or multiple genes. The detected interaction nodes will be annotated for downstream validation.

This package contain functions to run genomic instability analysis (GIA) from scRNA-Seq data. GIA estimates the association between gene expression and genomic location of the coding genes. It uses the aREA algorithm to quantify the enrichment of sets of contiguous genes (loci-blocks) on the gene expression profiles and estimates the Genomic Instability Score (GIS) for each analyzed cell.

Toolbox for comparative genomics of MAGs

Eukaryotic Genome Annotation Pipeline-External caller scripts and documentation

RFdiffusion is an open source method for structure generation, with or without conditional information (a motif, target etc).

toscca is an R package to perform Thresholded Ordered Sparse Canonical Correlation Analysis (TOSCCA).

Point mutations occurring in a genome can be divided into 96 categories based on the base being mutated, the base it is mutated into and its two flanking bases. Therefore, for any patient, it is possible to represent all the point mutations occurring in that patient's tumor as a vector of length 96, where each element represents the count of mutations for a given category in the patient. A mutational signature represents the pattern of mutations produced by a mutagen or mutagenic process inside the cell. Each signature can also be represented by a vector of length 96, where each element represents the probability that this particular mutagenic process generates a mutation of the 96 above mentioned categories. In this R package, we provide a set of functions to extract and visualize the mutational signatures that best explain the mutation counts of a large number of patients.

Cancer is a genetic disease caused by somatic mutations in genes controlling key biological functions such as cellular growth and division. Such mutations may arise both through cell-intrinsic and exogenous processes, generating characteristic mutational patterns over the genome named mutational signatures. The study of mutational signatures have become a standard component of modern genomics studies, since it can reveal which (environmental and endogenous) mutagenic processes are active in a tumor, and may highlight markers for therapeutic response. Mutational signatures computational analysis presents many pitfalls. First, the task of determining the number of signatures is very complex and depends on heuristics. Second, several signatures have no clear etiology, casting doubt on them being computational artifacts rather than due to mutagenic processes. Last, approaches for signatures assignment are greatly influenced by the set of signatures used for the analysis. To overcome these limitations, we developed RESOLVE (Robust EStimation Of mutationaL signatures Via rEgularization), a framework that allows the efficient extraction and assignment of mutational signatures. RESOLVE implements a novel algorithm that enables (i) the efficient extraction, (ii) exposure estimation, and (iii) confidence assessment during the computational inference of mutational signatures.

Single-cell RNA-seq technologies enable high throughput gene expression measurement of individual cells, and allow the discovery of heterogeneity within cell populations. Measurement of cell-to-cell gene expression similarity is critical for the identification, visualization and analysis of cell populations. However, single-cell data introduce challenges to conventional measures of gene expression similarity because of the high level of noise, outliers and dropouts. We develop a novel similarity-learning framework, SIMLR (Single-cell Interpretation via Multi-kernel LeaRning), which learns an appropriate distance metric from the data for dimension reduction, clustering and visualization.

LACE is an algorithmic framework that processes single-cell somatic mutation profiles from cancer samples collected at different time points and in distinct experimental settings, to produce longitudinal models of cancer evolution. The approach solves a Boolean Matrix Factorization problem with phylogenetic constraints, by maximizing a weighed likelihood function computed on multiple time points.

OncoScore is a tool to measure the association of genes to cancer based on citation frequencies in biomedical literature. The score is evaluated from PubMed literature by dynamically updatable web queries.

Detection of rare aberrant splicing events in transcriptome profiles. Read count ratio expectations are modeled by an autoencoder to control for confounding factors in the data. Given these expectations, the ratios are assumed to follow a beta-binomial distribution with a junction specific dispersion. Outlier events are then identified as read-count ratios that deviate significantly from this distribution. FRASER is able to detect alternative splicing, but also intron retention. The package aims to support diagnostics in the field of rare diseases where RNA-seq is performed to identify aberrant splicing defects.

Identification of aberrant gene expression in RNA-seq data. Read count expectations are modeled by an autoencoder to control for confounders in the data. Given these expectations, the RNA-seq read counts are assumed to follow a negative binomial distribution with a gene-specific dispersion. Outliers are then identified as read counts that significantly deviate from this distribution. Furthermore, OUTRIDER provides useful plotting functions to analyze and visualize the results.

SCENIC+ is a python package to build gene regulatory networks (GRNs) using combined or separate single-cell gene expression (scRNA-seq) and single-cell chromatin accessibility (scATAC-seq) data.

MitoFinder: efficient automated large-scale extraction of mitogenomic data from high throughput sequencing data

A database system designed to store, organize, and manage large-scale nucleotide sequencing read data (like PacBio reads) for the Dazzler genome assembler

Cis-mQTL mapping protocol for placental methylome using TensorQTL, providing a step-by-step guide for genotype QC, imputation, and analysis with detailed scripts and commands.

The hdxmsqc package enables us to analyse and visualise the quality of HDX-MS experiments. Either as a final quality check before downstream analysis and publication or as part of a interative procedure to determine the quality of the data. The package builds on the QFeatures and Spectra packages to integrate with other mass-spectrometry data.

Short Python script (using Biopython library functions) to extract sequences from a FASTA, QUAL, FASTQ, or SFF file based on the list of IDs given by a column of a tabular file. The output order follows that of the tabular file, and if there are duplicates in the tabular file, there will be duplicates in the output sequence file.

This package implements functions for finding breakpoints, plotting and export of Strand-seq data.

Fully automatic CAC scoring analysis of QRM CCI insert

The Connectivity Map (CMap) is a massive resource of perturbational gene expression profiles built by researchers at the Broad Institute and funded by the NIH Library of Integrated Network-Based Cellular Signatures (LINCS) program. Please visit https://clue.io for more information. The cmapR package implements methods to parse, manipulate, and write common CMap data objects, such as annotated matrices and collections of gene sets.

barcodetrackR is an R package developed for the analysis and visualization of clonal tracking data. Data required is samples and tag abundances in matrix form. Usually from cellular barcoding experiments, integration site retrieval analyses, or similar technologies.

cTTN is a set of R scripts for analyzing genetic constraint in the TTN gene by adjusting for sequencing depth and calculating constraint metrics with confidence intervals. It includes tools for obtaining observed and expected variant counts from gnomAD exome data.

CHETAH (CHaracterization of cEll Types Aided by Hierarchical classification) is an accurate, selective and fast scRNA-seq classifier. Classification is guided by a reference dataset, preferentially also a scRNA-seq dataset. By hierarchical clustering of the reference data, CHETAH creates a classification tree that enables a step-wise, top-to-bottom classification. Using a novel stopping rule, CHETAH classifies the input cells to the cell types of the references and to "intermediate types": more general classifications that ended in an intermediate node of the tree.

HyPPI classifies a protein-protein complex based on its interaction type into permanent, transient, or crystal artifact. Permanent protein-protein complexes are only stable in their complexed state. Their subunits would denature upon dissociation of the protein-protein complex. Transient protein-protein complexes are stable in the complexed as well as in the monomeric form, depending on the necessary function of the complex. Crystal artifacts have no biological function and are artificially formed during the crystallization process. The discrimination is performed using two characteristics of the protein-protein complex, the hydrophobicity of the interface (ΔGhydrophobic) and the quotient of interface area ratios (IF-quotient). The IF-quotient considers whether the protein-protein interface is symmetric.

JAMDA enables the preparation of individual protein structures and the docking of small molecules in preprocessed binding sites of choice. JAMDA simplifies the process of protein-ligand docking by automatic preprocessing protocols for the protein and binding sites of interest. The JAMDAscore scoring function retrieved 75% of the native poses in the three highest-ranked solutions for high-quality protein-ligand complexes with default settings. Individual configurations for protein preparation are available, e.g., considering protein ensembles, relevant binding site water molecules, or cofactors. A user-defined number of input conformations for the ligands of interest can be generated fully automated using Conformator. Alternatively, users can also provide externally prepared ligand conformers.

DoGSite3 was developed for predicting robust and reliable small molecule binding sites and computing their geometrical and chemical descriptors. It is based on the grid-based DoGSite algorithm for predicting pockets and their sub-pockets. The new tool is largely rotation- and translation-invariant due to a normalization procedure before binding site prediction. Known ligands in the structure can be used to bias the grid by sufficiently buried ligand fragments. The output encompasses novel chemical binding site descriptors considering solvent accessibility. Compared to its predecessor, it shows increased robustness through comprehensive parameter optimization. DoGSite3 runs finish within seconds.

DoGSiteScorer is a grid-based automated pocket detection and analysis tool. It applies a Difference of Gaussian filter to detect potential binding pockets and splits them into sub-pockets. The method solely uses the 3D structure of the protein. Global properties, describing the size, shape, and chemical features of the predicted (sub-)pockets, are calculated. Per default, a simple druggability score based on a linear combination of the three descriptors describing volume, hydrophobicity, and enclosure is provided for each (sub-)pocket. Furthermore, a subset of meaningful descriptors is incorporated in a support vector machine (libsvm) to predict the (sub-)pocket druggability score (values are between zero and one). The higher the score, the more druggable the pocket is estimated to be.

PoseEdit automatically generates 2D diagrams of protein-ligand complexes, focusing on the interactions between protein and ligand. Interactions between molecules are estimated by an underlying interaction model that relies on atom types and simple geometric criteria. The structure mining tool GeoMine also uses this model to describe binding sites. In addition, users can manipulate the diagrams by translating, rotating, mirroring parts of the structure, adding additional interactions, or removing them. Furthermore, users can add individual labels or adjust available labels. Users can download the final 2D diagrams for a binding site of interest in JSON or SVG format.

METALizer predicts the coordination geometry of metal ions in metalloproteins. Users can compare potential coordination geometries to those found in the examined structure. The predicted coordination geometries and the observed metal interaction distances can be interactively compared to statistics calculated based on the PDB.

PoseView automatically generates 2D diagrams of protein-ligand complexes, focusing on the interactions between protein and ligand. Interactions between molecules are estimated by an underlying interaction mode that relies on atom types and simple geometric criteria. It adheres to the conventions of chemical structure diagram generation. The quality of the resulting diagrams is comparable to manually drawn examples from books and scientific publications.

MicroMiner assists in identifying single-residue substitutions in protein structure databases. It searches protein residue environments with local sequence and structural similarity based on the SIENA methodology. Users can search for structural mutation in the entire PDB, their in-house structure collection, or (subsets of) the AlphaFold Database. They can use the method to explore the mutation landscape of proteins with experimental or predicted structures. MicroMiner can be applied to single domains or even protein-protein or protein-ligand interfaces. Several filter options to simplify downstream analysis are available.

SIENA is a software pipeline enabling the fully automated construction of protein structure ensembles from the PDB. Starting with a single query structure, all binding sites with high sequence similarity are extracted from the PDB, aligned, and superimposed. SIENA also handles complicated cases, such as comparing binding sites at protein domain interfaces or within multimeric proteins.

GeoMine enables the automated mining of protein-ligand binding sites. Based on individually designed queries, users can search for spatial interaction patterns in huge collections of protein-ligand complexes and binding pockets. The regularly updated GeoMine database relies on the free database systems SQLite and PostgreSQL. It supports radius-based pockets (based on ligands and predicted pockets (based on DoGSite3) for query generation. The query management is based on XML (for the REST service) or JSON in the GUI mode. Its output consists of the query-based superpositions of the matched binding sites and statistics on matching points, distances, and angles.

WarPP predicts the position and orientation of water molecules in small-molecule binding sites. It places and scores water molecules in binding sites of crystallographic structures based on EDIAscorer results and interaction geometries as known from experimentally solved protein structures. WarPP was validated on a high-quality set of 1,500 protein-ligand complexes, containing 20,000 crystallographically observed water molecules. It is sufficiently fast for high-throughput analyses. It correctly places water molecules in approx. 80% of the cases. Users can export the predictions as PDB files for, e.g., molecular docking with JAMDA.

Protoss is a fully automated hydrogen atom placement tool for protein-ligand complexes. It adds missing hydrogen atoms to protein structures and detects reasonable protonation states, tautomeric states, and hydrogen coordinates of both protein and ligand molecules by optimizing the hydrogen bond network.

The electron density score for individual atoms (EDIA) quantifies the electron density fit of each atom in a crystallographically resolved structure. Multiple EDIA values can be combined using the power mean to compute the EDIAm, i.e., the electron density score for a group of several atoms. It enables users to score a set of atoms, such as a ligand, a residue, or an active site.

Three-dimensional protein structures play a vital role in drug design. Structure-based design necessitates an in-depth examination of the available quality data before using the structure in computational experiments and for method evaluation. StructureProfiler assists in automatically profiling sets of protein-ligand complex structures based on multiple quality indicators, ranging from model characteristics, e.g., the R factor, and active site features, e.g., bond length deviations, to ligand properties such as electron density support and the validity of torsion angles.

LifeSoaks was designed to find solvent channels in macromolecular structures solved by X-ray crystallography. It predicts their accessibility by molecules through an automated annotation of so-called bottleneck radii. It simplifies the process of manually checking a crystal structure for solvent channels. Bottleneck radii can be calculated for solvent channels and small molecule binding sites. The tool is ideally suited for channel analyses before the actual soaking experiments to select the most promising experimental conditions and crystal forms. LifeSoaks runs fully automated and will finish within seconds to minutes for moderately sized crystals.

Normalizes a data matrix `data` by raking (using the RAS method by Bacharach, see references) the Nrows by Ncols matrix such that the row means and column means equal 1. The result is a normalized data matrix `K=RAS`, a product of row mulipliers `R` and column multipliers `S` with the original matrix `A`. Missing information needs to be presented as `NA` values and not as zero values, because CONSTANd is able to ignore missing values when calculating the mean. Using CONSTANd normalization allows for the direct comparison of values between samples within the same and even across different CONSTANd-normalized data matrices.

The package encompasses functions to find potential guide RNAs for the CRISPR-based genome-editing systems including the Base Editors and the Prime Editors when supplied with target sequences as input. Users have the flexibility to filter resulting guide RNAs based on parameters such as the absence of restriction enzyme cut sites or the lack of paired guide RNAs. The package also facilitates genome-wide exploration for off-targets, offering features to score and rank off-targets, retrieve flanking sequences, and indicate whether the hits are located within exon regions. All detected guide RNAs are annotated with the cumulative scores of the top5 and topN off-targets together with the detailed information such as mismatch sites and restrictuion enzyme cut sites. The package also outputs INDELs and their frequencies for Cas9 targeted sites.

DEMAND predicts Drug MoA by interrogating a cell context specific regulatory network with a small number (N >= 6) of compound-induced gene expression signatures, to elucidate specific proteins whose interactions in the network is dysregulated by the compound.

Inference of Genetic Variants Driving Cellullar Phenotypes by the DIGGIT algorithm

De novo identification and extraction of differentially methylated regions (DMRs) from the human genome using Whole Genome Bisulfite Sequencing (WGBS) and Illumina Infinium Array (450K and EPIC) data. Provides functionality for filtering probes possibly confounded by SNPs and cross-hybridisation. Includes GRanges generation and plotting functions.

The package allows one to obtain optimised combinations of DNA barcodes to be used for multiplex sequencing. In each barcode combination, barcodes are pooled with respect to Illumina chemistry constraints. Combinations can be filtered to keep those that are robust against substitution and insertion/deletion errors thereby facilitating the demultiplexing step. In addition, the package provides an optimiser function to further favor the selection of barcode combinations with least heterogeneity in barcode usage.