Initial commit: Chai-1 protein structure prediction pipeline for WES

- Nextflow pipeline using chai1 Docker image from Harbor
- S3-based input/output paths (s3://omic/eureka/chai-lab/)
- GPU-accelerated protein folding with MSA support

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

main.nf

@@ -0,0 +1,45 @@
+#!/usr/bin/env nextflow
+nextflow.enable.dsl=2
+
+params.input_dir = 's3://omic/eureka/chai-lab/input'
+params.outdir = 's3://omic/eureka/chai-lab/output'
+params.use_msa = true
+params.msa_server = 'https://api.colabfold.com'
+params.num_samples = 5
+
+process CHAI1 {
+    container 'harbor.cluster.omic.ai/omic/chai1:latest'
+    publishDir params.outdir, mode: 'copy'
+    stageInMode 'copy'
+    maxForks 1
+
+    input:
+        path fasta
+
+    output:
+        path "${fasta.simpleName.replace('.fasta', '')}", emit: output_dir
+
+    script:
+        """
+        OUTPUT_DIR=\$(basename ${fasta} .fasta)
+        mkdir -p \$OUTPUT_DIR
+
+        # Construct MSA parameters
+        MSA_OPTIONS=""
+        if ${params.use_msa}; then
+            MSA_OPTIONS="--use-msa-server --msa-server-url ${params.msa_server}"
+        fi
+
+        # Run CHAI1
+        chai fold \\
+            \$MSA_OPTIONS \\
+            --num-diffn-samples ${params.num_samples} \\
+            ${fasta} \\
+            \$OUTPUT_DIR
+        """
+}
+
+workflow {
+    fasta_ch = Channel.fromPath(params.input_dir + '/*.fasta')
+    CHAI1(fasta_ch)
+}