refactor: reconstruct examples/search file

Author: CHERRY-ui8

examples/search/build_db/build_protein_blast_db.sh

@@ -0,0 +1,84 @@
+# Search DNA Sequences
+
+This example demonstrates how to search DNA sequences from NCBI RefSeq database using BLAST.
+
+## Overview
+
+The DNA search pipeline reads DNA sequence queries and searches against NCBI RefSeq database to find similar sequences and retrieve associated metadata.
+
+## Quick Start
+
+### 1. Build Local BLAST Database (Optional)
+
+If you want to use local BLAST for faster searches, first build the database:
+
+```bash
+./build_db.sh [human_mouse_drosophila_yeast|representative|complete|all]
+```
+
+Options:
+- `human_mouse_drosophila_yeast`: Download only Homo sapiens, Mus musculus, Drosophila melanogaster, and Saccharomyces cerevisiae sequences (minimal, smallest)
+- `representative`: Download genomic sequences from major categories (recommended, smaller)
+- `complete`: Download all complete genomic sequences from complete/ directory (very large)
+- `all`: Download all genomic sequences from all categories (very large)
+
+The script will create a BLAST database in `refseq_${RELEASE}/` directory.
+
+### 2. Configure Search Parameters
+
+Edit `search_dna_config.yaml` to set:
+
+- **Input file path**: Set the path to your DNA sequence queries
+- **NCBI parameters**:
+  - `email`: Your email address (required by NCBI)
+  - `tool`: Tool name for NCBI API
+  - `use_local_blast`: Set to `true` if you have a local BLAST database
+  - `local_blast_db`: Path to your local BLAST database (without .nhr extension)
+
+Example configuration:
+```yaml
+input_path:
+  - examples/input_examples/search_dna_demo.jsonl
+
+data_sources: [ncbi]
+ncbi_params:
+  email: [email protected]  # Required!
+  tool: GraphGen
+  use_local_blast: true
+  local_blast_db: refseq_release/refseq_release
+```
+
+### 3. Run the Search
+
+```bash
+./search_dna.sh
+```
+
+Or run directly with Python:
+
+```bash
+python3 -m graphgen.run \
+  --config_file examples/search/search_dna/search_dna_config.yaml \
+  --output_dir cache/
+```
+
+## Input Format
+
+The input file should be in JSONL format with DNA sequence queries:
+
+```jsonl
+{"type": "text", "content": "BRCA1"}
+{"type": "text", "content": ">query\nATGCGATCG..."}
+{"type": "text", "content": "ATGCGATCG..."}
+```
+
+## Output
+
+The search results will be saved in the output directory with matched sequences and metadata from NCBI RefSeq.
+
+## Notes
+
+- **NCBI requires an email address** - Make sure to set `email` in `ncbi_params`
+- **Local BLAST** provides faster searches and doesn't require internet connection during search
+- The local BLAST database can be very large (several GB to TB depending on the download type)
+- Adjust `max_concurrent` based on your system resources and API rate limits