alloGlyph: transfer-entropy allostery alphabets (pooled + family PPI)

Large alphabet parquet files for the alloGlyph web server (transfer entropy allostery analysis). These three files exceed GitHub's 100 MB limit and are archived here; fetch_data.sh in the repository downloads them into data/. The v2 merged pooled alphabet (61.9M case-merged transfer-entropy patterns) is the default engine for both sequence and structure modes. The v1 merged alphabet and the per-family single-domain alphabet are provided for reproducibility. File Size (bytes) Role pooled_pattern_database_merged_v2.parquet 8,805,245,292 REQUIRED — default alphabet (61.9M patterns), all three modes pooled_pattern_database_merged.parquet 4,937,880,851 optional — v1 merged alphabet (33.9M patterns), fallback family_single_alphabet.parquet 1,310,534,403 optional — per-family single-domain alphabet (48.8M rows, 3,379 families); merged into v2, not used at runtime SHA-256 checksums pooled_pattern_database_merged_v2.parquet: d196a37110157e5f7c35cb1ea51a190e9f80bf13054a1986b9b6b1ffd6e3d4a9 pooled_pattern_database_merged.parquet: bc7f6ab366a35c8118cd77616144ff01253f04e4f749603b932e6426bb02a04f family_single_alphabet.parquet: 25a34f738fee95ddcd9ebfcc8845b50612cbd6cdd5b29fc24937d6753021103a Minimum to run: only the v2 file is required (8.2 GB); the other two are for reproducibility / fallback.