Align, classify, and humanise antibody sequences with Humatch in Docker.
npx skills add https://github.com/yjcyxky/biominer-skills --skill humatch
Please help me install this Agent Skill. Skill: humatch Source: https://github.com/yjcyxky/biominer-skills/tree/main/humatch Command: npx skills add https://github.com/yjcyxky/biominer-skills --skill humatch
This Skill automates the end-to-end processing of antibody sequences by performing alignment, humanness classification, and joint humanisation using Humatch inside a Docker container. It enables single-sequence workflows as well as batch processing via CSV inputs, with configurable germline-likeness constraints and CNN-based targeting for quality humanisation.
Start the Humatch workflow by running the container image with your data (single pair or CSV) and following the examples in the SKILL description. For example, align a heavy and light chain or run a full classification and humanisation pipeline using the provided CLI.
High-intent search queries and answers about installing and using this skill.
Antibody humanisation and alignment for VH and VL chains is executed via CLI inside a Docker container using the Humatch image. You process heavy and light chains individually or together with configurable germline-likeness targets.
Yes, batch processing antibody sequences is supported using CSV inputs. The Docker-based Humatch workflow reads the CSV file to align sequences, classify humanness, and generate humanised outputs for multiple candidates.
Docker is required for humanness classification and germline-likeness targeting. The Skill runs end-to-end processing inside the Humatch container image, ensuring reproducible alignment and humanisation under CNN constraints.
Antibody humanisation modifies non-human antibody sequences to increase human likeness. Prealigned 200-position inputs are needed when processing preformatted heavy or light chain sequences for CNN-based humanness classification.
The Humatch Docker container requires network access for initial model asset downloads. If auto-download fails due to network restrictions, you can manually supply the necessary model assets to the container.
Yes, fixed target positions for antibody humanisation can be configured using a YAML config file. This allows tuning behavior and setting specific germline-likeness constraints during the CNN-based humanisation process.
pagopaCreate and modify Dockerfiles with digest-pinned images and secure runtime defaults.
PharMolixPredict antibody and nanobody structures from sequences using tfold-based models.
wuyoscarAutomate toxin protein design and scoring with PyRosetta pipelines.
cliffthelinExtend Docker environments with gitignored override files and apt-get tool installation.
