AmirAli
Toghani

Google Colab notebooks providing single-sequence and high-throughput batch structure prediction powered by ByteDance's Protenix-v2, an open-source all-atom biomolecular modeling system. Predicts structures of proteins, DNA, RNA, and small-molecule ligand complexes.

Google Colab notebooks providing single-run and batch modeling via OpenDDE (Open-Source Drug Design Engine) by Aureka Research, tailored for macromolecular complex modeling, binding interface prediction, and structure-based discovery workflows.

Google Colab pipeline implementing RFdiffusion3 for de novo design of high-affinity protein binders against specified molecular targets. Generates complementary binder backbones conditioned on target structures with full atom-level accuracy.

Google Colab pipeline for end-to-end de novo protein binder design powered by BindCraft 2 (BC2). Drives AlphaFold 2 gradient-based backbone design (hallucination), ProteinMPNN sequence redesign, held-out model re-prediction, automated multi-metric filtering, and ranking by interface confidence (i_pDAE) across diverse modalities and multi-target campaigns.

Google Colab notebook deploying OrionGeno for accurate ab initio eukaryotic gene prediction directly from genomic FASTA sequences. Utilizes modern deep learning to resolve exon-intron boundaries and complex genomic architecture.

Google Colab notebook running Tiberius, an end-to-end gene prediction framework that couples convolutional neural networks and transformer layers with a differentiable Hidden Markov Model (HMM) for rapid, accurate eukaryotic genome annotation.

Google Colab notebook deploying Helixer for fast, high-accuracy ab initio eukaryotic gene prediction coupled with automated BUSCO completeness assessment. Integrates convolutional and bidirectional LSTM networks with HMM decoding (HelixerPost), multi-lineage model support, and gffread CDS/protein extraction across single or batch genomes.

Google Colab notebook deploying Carbon-A-1.2B (Hugging Face Biology), a 1.2-billion-parameter foundation DNA language model that predicts base-wise protein-coding probabilities across eukaryotic genomes. Combines a gene-grammar Viterbi decoder to resolve full gene models (GFF3, CDS, and protein sequences), per-strand bigWig coding-probability tracks, and automated BUSCO completeness assessment for single or batch genomes.

Google Colab notebook deploying BUSCO (Benchmarking Universal Single-Copy Orthologs) for automated completeness assessment of proteomes, genome assemblies, and transcriptomes against OrthoDB lineages. Supports single files, batches, and Google Drive folders across proteins (HMMER), genome (miniprot/MetaEuk), and transcriptome modes, generating consolidated summary tables (TSV), publication-ready stacked-bar plots, and sample status matrices.

An interactive planner for Golden Gate and Type IIS restriction-ligation cloning. Calculates exact pipetting volumes for plasmid preps based on measured concentrations and molar ratios relative to the destination vector.

Features automated Type IIS recognition site auditing across circular sequences, water make-up calculations, shared master mix pooling with configurable excess, part inventory tracking across multi-tube runs, and a quick dilution calculator (C₁·V₁ = C₂·V₂).

An interactive bench tool designed for bacterial dilution planning and OD₆₀₀ normalization. Calculates exact neat culture and buffer volumes needed to normalize bacterial suspensions to target densities.

Supports single-strain dilutions and multi-strain co-culture combinations, tracks construct metadata (strains, plasmids, tags, antibiotics, stock locations), accounts for pipetting dead volume, and exports formatted run sheets for the bench.