Implements, modifies, or debugs constraints in the proto-language DSL. Covers the full lifecycle: BaseConfig class with ConfigField, scoring function returning list[ConstraintOutput], @constraint decorator registration, 3-level export chain, and pytest test coverage. Use when working with constraints, scoring functions, GC content, structure prediction scores (pLDDT, pTM, pAE), protein quality, sequence motifs, RNA structure, or splicing predictions.
Implements, modifies, or debugs generators in the proto-language DSL. Covers the full lifecycle: config class, Generator subclass with __init__/assign/sample, input_type-specific patterns (mutation, autoregressive, inverse folding, gradient), decorator registration, export chain, and tests. Use when working with generators, sequence sampling, masked LMs, causal LMs, inverse folding, or mutation strategies.
Implements, modifies, or debugs optimizers in the proto-language DSL. Covers the full lifecycle: BaseOptimizerConfig with ConfigField, Optimizer subclass with __init__/run, dual-pool architecture (result/proposal sequences), constraint evaluation (filter + scoring), decorator registration, export chain, and pytest test coverage. Use when working with optimizers, MCMC, beam search, rejection sampling, cycling, or sequence optimization algorithms.
Composes proto-language optimization programs in Python. Covers Segments, Constructs, Generators, Constraints, Optimizers, and Programs for designing biological sequences. Use when writing programs, composing optimization pipelines, designing DNA/protein/RNA sequences, or setting up multi-stage optimization with constraints like GC content, structure prediction, or protein quality.