Add STRING WaveGC and STRING GNN to catalog

catalog.json

@@ -1,4 +1,48 @@
 [
+  {
+    "algorithm": "AAC.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 93.91,
+    "dataset_path": "data/embeddings/probe_aac",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_aac",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_aac/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 20,
+    "n_genes": 18873,
+    "name": "aac",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_aac",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
+  {
+    "algorithm": "ALBERT.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 91.82,
+    "dataset_path": "data/embeddings/probe_albert",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_albert",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_albert/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 4096,
+    "n_genes": 18453,
+    "name": "albert",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_albert",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
   {
     "algorithm": "Google DeepMind AlphaGenome API regulatory predictions. signed log1p per summary feature, z-score across genes, PCA to 256 dimensions.",
     "category": "Regulatory",
@@ -21,6 +65,28 @@
     "zero_vectors": 0,
     "license_url": "https://github.com/google-deepmind/alphagenome"
   },
+  {
+    "algorithm": "APAAC.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 93.9,
+    "dataset_path": "data/embeddings/probe_apaac",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_apaac",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_apaac/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 80,
+    "n_genes": 18871,
+    "name": "apaac",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_apaac",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
   {
     "algorithm": "Basenji2 sequence-to-function model (checkpoint: Basenji2 PyTorch port). mean of central 3 output bins around TSS. log1p raw human central3 mean tracks, then dimension-wise z-score across genes.",
     "category": "Regulatory",
@@ -43,6 +109,50 @@
     "zero_vectors": 0,
     "license_url": "https://zenodo.org/records/14969593"
   },
+  {
+    "algorithm": "BERT-BFD.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 91.82,
+    "dataset_path": "data/embeddings/probe_bert_bfd",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_bert_bfd",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_bert_bfd/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 1024,
+    "n_genes": 18453,
+    "name": "bert_bfd",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_bert_bfd",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
+  {
+    "algorithm": "BERT-PFAM.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 91.86,
+    "dataset_path": "data/embeddings/probe_bert_pfam",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_bert_pfam",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_bert_pfam/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 768,
+    "n_genes": 18461,
+    "name": "bert_pfam",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_bert_pfam",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
   {
     "algorithm": "BioBERT-GeneProteinSummary (checkpoint: dmis-lab/biobert-base-cased-v1.2). final hidden layer mean pooling over valid non-special, non-padding tokens.",
     "category": "Literature",
@@ -153,6 +263,28 @@
     "zero_vectors": 0,
     "license_url": "https://github.com/ncbi/BioConceptVec"
   },
+  {
+    "algorithm": "BLAST.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 93.91,
+    "dataset_path": "data/embeddings/probe_blast",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_blast",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_blast/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 20421,
+    "n_genes": 18873,
+    "name": "blast",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_blast",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
   {
     "algorithm": "CMap/LINCS L1000 consensus gene signatures. mean 978-dimensional L1000 response vector across signatures and cell lines per perturbed gene. GCTX matrix interpreted as CGS signatures x 978 landmark response genes. Signatures were filtered to pert_type trt_sh.cgs, averaged by perturbed gene symbol across cell lines/signatures, then landmark response gene columns were z-scored across mapped perturbed genes.",
     "category": "Expression",
@@ -175,6 +307,28 @@
     "zero_vectors": 0,
     "license_url": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE106127"
   },
+  {
+    "algorithm": "CPC-PROT.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 91.82,
+    "dataset_path": "data/embeddings/probe_cpc_prot",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_cpc_prot",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_cpc_prot/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 512,
+    "n_genes": 18453,
+    "name": "cpc_prot",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_cpc_prot",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
   {
     "algorithm": "deepNF: deep network fusion for protein function prediction.",
     "category": "Network",
@@ -307,6 +461,28 @@
     "zero_vectors": 0,
     "license_url": "https://zenodo.org/records/16764517"
   },
+  {
+    "algorithm": "ESMB1.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 91.82,
+    "dataset_path": "data/embeddings/probe_esmb1",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_esmb1",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_esmb1/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 1280,
+    "n_genes": 18453,
+    "name": "esmb1",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_esmb1",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
   {
     "algorithm": "FRoGS-ARCHS4-256 (checkpoint: FRoGS ARCHS4 gene embedding).",
     "category": "Functional Genomics",
@@ -351,6 +527,28 @@
     "zero_vectors": 0,
     "license_url": "https://github.com/jingcheng-du/Gene2vec"
   },
+  {
+    "algorithm": "GENE2VEC.",
+    "category": "Multimodal",
+    "category_source": "manual",
+    "community_contribution": false,
+    "coverage_pct": 87.4,
+    "dataset_path": "data/embeddings/probe_gene2vec_uniprot",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_gene2vec_uniprot",
+    "license": "PROBE source / license information",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_gene2vec_uniprot/metadata.json",
+    "modality": "multimodal",
+    "n_dims": 200,
+    "n_genes": 17564,
+    "name": "gene2vec_uniprot",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_gene2vec_uniprot",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://github.com/tuncadogan/PROBE"
+  },
   {
     "algorithm": "GeneCompass Base, 12-layer knowledge-informed cross-species foundation model. none; static learned token embedding weight extracted directly from checkpoint.",
     "category": "Single-cell",
@@ -505,6 +703,28 @@
     "zero_vectors": 0,
     "license_url": "https://zenodo.org/records/268568"
   },
+  {
+    "algorithm": "HMMER.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 93.91,
+    "dataset_path": "data/embeddings/probe_hmmer",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_hmmer",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_hmmer/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 20421,
+    "n_genes": 18873,
+    "name": "hmmer",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_hmmer",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
   {
     "algorithm": "HPA normal tissue immunohistochemistry annotations. maximum ordinal protein-expression level across annotated cell types within each tissue. HPA Level encoded as Not detected=0, Low=1, Medium=2, High=3. For each gene and tissue, maximum level across annotated cell types was used. Tissue columns were z-scored across mapped genes.",
     "category": "Protein",
@@ -635,32 +855,76 @@
     "license_url": "https://huggingface.co/LongSafari/hyenadna-medium-160k-seqlen-hf"
   },
   {
-    "algorithm": "Mahi context-aware gene embedding model. mean across all non-global tissue/cell contexts per Entrez gene.",
-    "category": "Multimodal",
-    "category_source": "manual",
+    "algorithm": "KSEP.",
+    "category": "Protein",
+    "category_source": "heuristic",
     "community_contribution": false,
-    "coverage_pct": 94.13,
-    "dataset_path": "data/embeddings/mahi_all_contexts_mean",
-    "description": "Mahi gene representation formed by averaging embeddings across all available non-global tissue and cell contexts.",
-    "id": "mahi_all_contexts_mean",
+    "coverage_pct": 93.55,
+    "dataset_path": "data/embeddings/probe_ksep",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_ksep",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/mahi_all_contexts_mean/metadata.json",
-    "modality": "multimodal tissue-aware graph representation / mean across non-global contexts",
-    "n_dims": 3666,
-    "n_genes": 18918,
-    "name": "mahi_all_contexts_mean",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/mahi_all_contexts_mean",
-    "qc_status": "processed_qc_passed",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_ksep/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 400,
+    "n_genes": 18801,
+    "name": "ksep",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_ksep",
+    "qc_status": "legacy_core_package",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://huggingface.co/datasets/anushaggs/Mahi_gene_embeddings"
+    "zero_vectors": 23,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
   },
   {
-    "algorithm": "Mahi context-aware gene embedding model. global context only.",
-    "category": "Multimodal",
-    "category_source": "manual",
-    "community_contribution": false,
+    "algorithm": "LEARNED-VEC.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 93.91,
+    "dataset_path": "data/embeddings/probe_learned_vec",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_learned_vec",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_learned_vec/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 64,
+    "n_genes": 18873,
+    "name": "learned_vec",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_learned_vec",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
+  {
+    "algorithm": "Mahi context-aware gene embedding model. mean across all non-global tissue/cell contexts per Entrez gene.",
+    "category": "Multimodal",
+    "category_source": "manual",
+    "community_contribution": false,
+    "coverage_pct": 94.13,
+    "dataset_path": "data/embeddings/mahi_all_contexts_mean",
+    "description": "Mahi gene representation formed by averaging embeddings across all available non-global tissue and cell contexts.",
+    "id": "mahi_all_contexts_mean",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/mahi_all_contexts_mean/metadata.json",
+    "modality": "multimodal tissue-aware graph representation / mean across non-global contexts",
+    "n_dims": 3666,
+    "n_genes": 18918,
+    "name": "mahi_all_contexts_mean",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/mahi_all_contexts_mean",
+    "qc_status": "processed_qc_passed",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/datasets/anushaggs/Mahi_gene_embeddings"
+  },
+  {
+    "algorithm": "Mahi context-aware gene embedding model. global context only.",
+    "category": "Multimodal",
+    "category_source": "manual",
+    "community_contribution": false,
     "coverage_pct": 94.13,
     "dataset_path": "data/embeddings/mahi_global",
     "description": "Global-context Mahi gene representation integrating regulatory, protein and network-derived molecular information.",
@@ -722,6 +986,28 @@
     "zero_vectors": 0,
     "license_url": "https://github.com/facebookresearch/esm"
   },
+  {
+    "algorithm": "MUT2VEC.",
+    "category": "Multimodal",
+    "category_source": "manual",
+    "community_contribution": false,
+    "coverage_pct": 86.8,
+    "dataset_path": "data/embeddings/probe_mut2vec",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_mut2vec",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_mut2vec/metadata.json",
+    "modality": "multimodal",
+    "n_dims": 300,
+    "n_genes": 17445,
+    "name": "mut2vec",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_mut2vec",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
   {
     "algorithm": "NEWT component embedding (checkpoint: not applicable). precomputed component embedding table.",
     "category": "Multimodal",
@@ -920,6 +1206,28 @@
     "zero_vectors": 0,
     "license_url": "https://huggingface.co/quietflamingo/orthrus-base-4-track"
   },
+  {
+    "algorithm": "PFAM.",
+    "category": "Protein",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 93.91,
+    "dataset_path": "data/embeddings/probe_pfam",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "probe_pfam",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_pfam/metadata.json",
+    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
+    "n_dims": 6227,
+    "n_genes": 18873,
+    "name": "pfam",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_pfam",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 1532,
+    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+  },
   {
     "algorithm": "PrimeKG heterogeneous biomedical knowledge graph. Built sparse gene x typed-neighbor-feature count matrix; collapsed duplicate gene-feature edges by summation; applied sublinear TF-IDF weighting; reduced with randomized TruncatedSVD; z-scored SVD dimensions across source genes; mapped to master by Entrez ID.",
     "category": "Knowledge",
@@ -943,21 +1251,21 @@
     "license_url": "https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/IXA7BM"
   },
   {
-    "algorithm": "AAC.",
+    "algorithm": "PROTVEC.",
     "category": "Protein",
     "category_source": "heuristic",
     "community_contribution": false,
     "coverage_pct": 93.91,
-    "dataset_path": "data/embeddings/probe_aac",
+    "dataset_path": "data/embeddings/probe_protvec",
     "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_aac",
+    "id": "probe_protvec",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_aac/metadata.json",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_protvec/metadata.json",
     "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 20,
+    "n_dims": 100,
     "n_genes": 18873,
-    "name": "aac",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_aac",
+    "name": "protvec",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_protvec",
     "qc_status": "legacy_core_package",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
@@ -965,197 +1273,153 @@
     "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
   },
   {
-    "algorithm": "ALBERT.",
-    "category": "Protein",
-    "category_source": "heuristic",
+    "algorithm": "PubMedBERT-GeneProteinSummary (checkpoint: microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext). final hidden layer mean pooling over valid non-special, non-padding tokens.",
+    "category": "Literature",
+    "category_source": "manual",
     "community_contribution": false,
-    "coverage_pct": 91.82,
-    "dataset_path": "data/embeddings/probe_albert",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_albert",
+    "coverage_pct": 95.38,
+    "dataset_path": "data/embeddings/pubmedbert_gene_protein_summary",
+    "description": "NCBI gene card summaries plus UniProt protein summaries when available",
+    "id": "pubmedbert_gene_protein_summary",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_albert/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 4096,
-    "n_genes": 18453,
-    "name": "albert",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_albert",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/pubmedbert_gene_protein_summary/metadata.json",
+    "modality": "literature_text_gene_summary",
+    "n_dims": 768,
+    "n_genes": 19168,
+    "name": "PubMedBERT-GeneProteinSummary",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/pubmedbert_gene_protein_summary",
     "qc_status": "legacy_core_package",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
     "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+    "license_url": "https://huggingface.co/microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext"
   },
   {
-    "algorithm": "APAAC.",
-    "category": "Protein",
-    "category_source": "heuristic",
+    "algorithm": "RNA-FM / rna_fm_t12 RNA language model. Final layer 12 mean token pooling over real RNA tokens only; BOS/EOS excluded; long transcripts split into chunks and chunk means length-weighted.",
+    "category": "RNA",
+    "category_source": "manual",
     "community_contribution": false,
-    "coverage_pct": 93.9,
-    "dataset_path": "data/embeddings/probe_apaac",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_apaac",
+    "coverage_pct": 99.4,
+    "dataset_path": "data/embeddings/rnafm_full_transcript",
+    "description": "Mature-transcript representation generated from the final layer of RNA-FM using length-weighted token pooling.",
+    "id": "rnafm_full_transcript",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_apaac/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 80,
-    "n_genes": 18871,
-    "name": "apaac",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_apaac",
-    "qc_status": "legacy_core_package",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/rnafm_full_transcript/metadata.json",
+    "modality": "RNA language model / mature mRNA transcript",
+    "n_dims": 640,
+    "n_genes": 19976,
+    "name": "rnafm_full_transcript",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/rnafm_full_transcript",
+    "qc_status": "processed_qc_passed",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
     "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+    "license_url": "https://github.com/ml4bio/RNA-FM"
   },
   {
-    "algorithm": "BERT-BFD.",
-    "category": "Protein",
+    "algorithm": "scFoundation / xTrimoGene (checkpoint: models.ckpt). fixed checkpoint embedding table; no cell/context pooling.",
+    "category": "Single-cell",
     "category_source": "heuristic",
     "community_contribution": false,
-    "coverage_pct": 91.82,
-    "dataset_path": "data/embeddings/probe_bert_bfd",
+    "coverage_pct": 94.85,
+    "dataset_path": "data/embeddings/scfoundation_gene_pos",
     "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_bert_bfd",
+    "id": "scfoundation_gene_pos",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_bert_bfd/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 1024,
-    "n_genes": 18453,
-    "name": "bert_bfd",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_bert_bfd",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scfoundation_gene_pos/metadata.json",
+    "modality": "single_cell_expression_foundation_model / learned gene identity-position embedding",
+    "n_dims": 768,
+    "n_genes": 19062,
+    "name": "scFoundation-gene-pos",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scfoundation_gene_pos",
     "qc_status": "legacy_core_package",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
     "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+    "license_url": "https://github.com/biomap-research/scFoundation"
   },
   {
-    "algorithm": "BERT-PFAM.",
-    "category": "Protein",
+    "algorithm": "scGPT whole-human input gene-token embedding layer. none; static learned token embedding weight extracted directly from checkpoint.",
+    "category": "Single-cell",
     "category_source": "heuristic",
     "community_contribution": false,
-    "coverage_pct": 91.86,
-    "dataset_path": "data/embeddings/probe_bert_pfam",
+    "coverage_pct": 95.93,
+    "dataset_path": "data/embeddings/scgpt_whole_human",
     "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_bert_pfam",
+    "id": "scgpt_whole_human",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_bert_pfam/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 768,
-    "n_genes": 18461,
-    "name": "bert_pfam",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_bert_pfam",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scgpt_whole_human/metadata.json",
+    "modality": "single_cell_expression_foundation_model",
+    "n_dims": 512,
+    "n_genes": 19280,
+    "name": "scGPT-whole-human",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scgpt_whole_human",
     "qc_status": "legacy_core_package",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
     "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+    "license_url": "https://github.com/bowang-lab/scGPT"
   },
   {
-    "algorithm": "BLAST.",
-    "category": "Protein",
-    "category_source": "heuristic",
+    "algorithm": "scGPT pan-cancer checkpoint (checkpoint: scGPT pan-cancer gene-token embedding). none; static learned token embedding weight extracted directly from checkpoint.",
+    "category": "Single-cell",
+    "category_source": "manual",
     "community_contribution": false,
-    "coverage_pct": 93.91,
-    "dataset_path": "data/embeddings/probe_blast",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_blast",
+    "coverage_pct": 94.89,
+    "dataset_path": "data/embeddings/scgpt_pancancer",
+    "description": "Gene-token representation derived from a pan-cancer scGPT single-cell transcriptomic model.",
+    "id": "scgpt_pancancer",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_blast/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 20421,
-    "n_genes": 18873,
-    "name": "blast",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_blast",
-    "qc_status": "legacy_core_package",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scgpt_pancancer/metadata.json",
+    "modality": "single-cell transcriptomic foundation model / pan-cancer gene-token embedding",
+    "n_dims": 512,
+    "n_genes": 19071,
+    "name": "scgpt_pancancer",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scgpt_pancancer",
+    "qc_status": "processed_qc_passed",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
     "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+    "license_url": "https://github.com/bowang-lab/scGPT"
   },
   {
-    "algorithm": "CPC-PROT.",
-    "category": "Protein",
+    "algorithm": "scPRINT (checkpoint: medium-v1.5.ckpt). fixed checkpoint embedding table; no cell/context pooling.",
+    "category": "Single-cell",
     "category_source": "heuristic",
     "community_contribution": false,
-    "coverage_pct": 91.82,
-    "dataset_path": "data/embeddings/probe_cpc_prot",
+    "coverage_pct": 99.49,
+    "dataset_path": "data/embeddings/scprint_medium_v1_5_gene",
     "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_cpc_prot",
+    "id": "scprint_medium_v1_5_gene",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_cpc_prot/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 512,
-    "n_genes": 18453,
-    "name": "cpc_prot",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_cpc_prot",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
-  },
-  {
-    "algorithm": "ESMB1.",
-    "category": "Protein",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 91.82,
-    "dataset_path": "data/embeddings/probe_esmb1",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_esmb1",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_esmb1/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 1280,
-    "n_genes": 18453,
-    "name": "esmb1",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_esmb1",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
-  },
-  {
-    "algorithm": "GENE2VEC.",
-    "category": "Multimodal",
-    "category_source": "manual",
-    "community_contribution": false,
-    "coverage_pct": 87.4,
-    "dataset_path": "data/embeddings/probe_gene2vec_uniprot",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_gene2vec_uniprot",
-    "license": "PROBE source / license information",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_gene2vec_uniprot/metadata.json",
-    "modality": "multimodal",
-    "n_dims": 200,
-    "n_genes": 17564,
-    "name": "gene2vec_uniprot",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_gene2vec_uniprot",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scprint_medium_v1_5_gene/metadata.json",
+    "modality": "single_cell_expression_foundation_model / learned gene-token embedding",
+    "n_dims": 256,
+    "n_genes": 19995,
+    "name": "scPRINT-medium-v1.5-gene",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scprint_medium_v1_5_gene",
     "qc_status": "legacy_core_package",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
     "zero_vectors": 0,
-    "license_url": "https://github.com/tuncadogan/PROBE"
+    "license_url": "https://huggingface.co/jkobject/scPRINT"
   },
   {
-    "algorithm": "HMMER.",
+    "algorithm": "SEQVEC.",
     "category": "Protein",
     "category_source": "heuristic",
     "community_contribution": false,
     "coverage_pct": 93.91,
-    "dataset_path": "data/embeddings/probe_hmmer",
+    "dataset_path": "data/embeddings/probe_seqvec",
     "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_hmmer",
+    "id": "probe_seqvec",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_hmmer/metadata.json",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_seqvec/metadata.json",
     "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 20421,
+    "n_dims": 1024,
     "n_genes": 18873,
-    "name": "hmmer",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_hmmer",
+    "name": "seqvec",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_seqvec",
     "qc_status": "legacy_core_package",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
@@ -1163,136 +1427,72 @@
     "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
   },
   {
-    "algorithm": "KSEP.",
-    "category": "Protein",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 93.55,
-    "dataset_path": "data/embeddings/probe_ksep",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_ksep",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_ksep/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 400,
-    "n_genes": 18801,
-    "name": "ksep",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_ksep",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 23,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
-  },
-  {
-    "algorithm": "LEARNED-VEC.",
-    "category": "Protein",
+    "algorithm": "SpliceAI splice-site prediction model. aggregate canonical splice-site profiles to one fixed-size vector per gene.",
+    "category": "Regulatory",
     "category_source": "heuristic",
     "community_contribution": false,
-    "coverage_pct": 93.91,
-    "dataset_path": "data/embeddings/probe_learned_vec",
+    "coverage_pct": 100.0,
+    "dataset_path": "data/embeddings/spliceai_canonical_splice_site_profile",
     "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_learned_vec",
+    "id": "spliceai_canonical_splice_site_profile",
     "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_learned_vec/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 64,
-    "n_genes": 18873,
-    "name": "learned_vec",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_learned_vec",
-    "qc_status": "legacy_core_package",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/spliceai_canonical_splice_site_profile/metadata.json",
+    "modality": "splicing sequence model / canonical transcript splice-site profiles",
+    "n_dims": 606,
+    "n_genes": 20097,
+    "name": "SpliceAI-canonical-splice-site-profile-zscore",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/spliceai_canonical_splice_site_profile",
+    "qc_status": "processed_qc_passed",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
     "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+    "license_url": "https://github.com/Illumina/SpliceAI"
   },
   {
-    "algorithm": "MUT2VEC.",
-    "category": "Multimodal",
+    "id": "string_gnn",
+    "name": "STRING GNN",
+    "category": "Network",
     "category_source": "manual",
-    "community_contribution": false,
-    "coverage_pct": 86.8,
-    "dataset_path": "data/embeddings/probe_mut2vec",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_mut2vec",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_mut2vec/metadata.json",
-    "modality": "multimodal",
-    "n_dims": 300,
-    "n_genes": 17445,
-    "name": "mut2vec",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_mut2vec",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "modality": "STRINGdb protein-protein association network",
+    "algorithm": "Residual graph convolutional network link prediction",
+    "description": "Human gene-level embeddings learned by GenBio AI from a filtered STRING v12.0 protein-protein association graph with a residual GCN and link-prediction objective.",
+    "n_genes": 18028,
+    "n_dims": 256,
+    "coverage_pct": 89.7,
+    "qc_status": "PASS",
+    "license": "GenBio AI Community License Agreement",
+    "license_url": "https://huggingface.co/datasets/genbio-ai/foundation-models-perturbation/blob/d67ba21019814bb55dbdb448b3b84d18bbc693b8/LICENSE",
     "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
-  },
-  {
-    "algorithm": "PFAM.",
-    "category": "Protein",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 93.91,
-    "dataset_path": "data/embeddings/probe_pfam",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_pfam",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_pfam/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 6227,
-    "n_genes": 18873,
-    "name": "pfam",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_pfam",
-    "qc_status": "legacy_core_package",
+    "dataset_path": "data/embeddings/string_gnn",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/f703e997fa18de926e750ff9ec4ad452cbdd8184/data/embeddings/string_gnn",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/f703e997fa18de926e750ff9ec4ad452cbdd8184/data/embeddings/string_gnn/metadata.json",
+    "source_url": "https://huggingface.co/datasets/genbio-ai/foundation-models-perturbation/blob/d67ba21019814bb55dbdb448b3b84d18bbc693b8/gene_embeddings/GNN_Simple_Official_%28D%3D256%29.h5ad",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 1532,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+    "source_revision": "f703e997fa18de926e750ff9ec4ad452cbdd8184",
+    "community_contribution": false
   },
   {
-    "algorithm": "PROTVEC.",
-    "category": "Protein",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 93.91,
-    "dataset_path": "data/embeddings/probe_protvec",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_protvec",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_protvec/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 100,
-    "n_genes": 18873,
-    "name": "protvec",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_protvec",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "id": "string_wavegc",
+    "name": "STRING WaveGC",
+    "category": "Network",
+    "category_source": "manual",
+    "modality": "STRINGdb protein-protein association network",
+    "algorithm": "WaveGC graph spectral wavelet convolution",
+    "description": "Human gene-level embeddings learned from the STRING protein-protein association network with WaveGC using a link-prediction objective.",
+    "n_genes": 18131,
+    "n_dims": 128,
+    "coverage_pct": 90.22,
+    "qc_status": "PASS",
+    "license": "GenBio AI Community License Agreement",
+    "license_url": "https://huggingface.co/datasets/genbio-ai/foundation-models-perturbation/blob/d67ba21019814bb55dbdb448b3b84d18bbc693b8/LICENSE",
     "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
-  },
-  {
-    "algorithm": "SEQVEC.",
-    "category": "Protein",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 93.91,
-    "dataset_path": "data/embeddings/probe_seqvec",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "probe_seqvec",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_seqvec/metadata.json",
-    "modality": "protein sequence / protein representation from PROBE human UniProt archive",
-    "n_dims": 1024,
-    "n_genes": 18873,
-    "name": "seqvec",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/probe_seqvec",
-    "qc_status": "legacy_core_package",
+    "dataset_path": "data/embeddings/string_wavegc",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/f703e997fa18de926e750ff9ec4ad452cbdd8184/data/embeddings/string_wavegc",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/f703e997fa18de926e750ff9ec4ad452cbdd8184/data/embeddings/string_wavegc/metadata.json",
+    "source_url": "https://huggingface.co/datasets/genbio-ai/foundation-models-perturbation/blob/d67ba21019814bb55dbdb448b3b84d18bbc693b8/gene_embeddings/KG_WaveGC_G_StringDB_combined_0.0_%28D%3D128%29.h5ad",
     "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
+    "source_revision": "f703e997fa18de926e750ff9ec4ad452cbdd8184",
+    "community_contribution": false
   },
   {
     "algorithm": "T5.",
@@ -1338,6 +1538,28 @@
     "zero_vectors": 0,
     "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
   },
+  {
+    "algorithm": "Universal Cell Embeddings (UCE) (checkpoint: 33l_8ep_1024t_1280.torch). UCE gene encoder projection: Linear(5120->1280), ReLU, LayerNorm.",
+    "category": "Single-cell",
+    "category_source": "heuristic",
+    "community_contribution": false,
+    "coverage_pct": 94.95,
+    "dataset_path": "data/embeddings/uce_33l",
+    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
+    "id": "uce_33l",
+    "license": "Original source / license terms",
+    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/uce_33l/metadata.json",
+    "modality": "single_cell_expression_foundation_model / protein-informed gene-token embedding",
+    "n_dims": 1280,
+    "n_genes": 19082,
+    "name": "UCE-33L",
+    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/uce_33l",
+    "qc_status": "legacy_core_package",
+    "source_dataset": "embeddingsSSRF/gene_embeddings",
+    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
+    "zero_vectors": 0,
+    "license_url": "https://github.com/snap-stanford/UCE"
+  },
   {
     "algorithm": "UNIREP.",
     "category": "Protein",
@@ -1381,181 +1603,5 @@
     "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
     "zero_vectors": 0,
     "license_url": "https://huggingface.co/spaces/HUBioDataLab/PROBE"
-  },
-  {
-    "algorithm": "PubMedBERT-GeneProteinSummary (checkpoint: microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext). final hidden layer mean pooling over valid non-special, non-padding tokens.",
-    "category": "Literature",
-    "category_source": "manual",
-    "community_contribution": false,
-    "coverage_pct": 95.38,
-    "dataset_path": "data/embeddings/pubmedbert_gene_protein_summary",
-    "description": "NCBI gene card summaries plus UniProt protein summaries when available",
-    "id": "pubmedbert_gene_protein_summary",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/pubmedbert_gene_protein_summary/metadata.json",
-    "modality": "literature_text_gene_summary",
-    "n_dims": 768,
-    "n_genes": 19168,
-    "name": "PubMedBERT-GeneProteinSummary",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/pubmedbert_gene_protein_summary",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://huggingface.co/microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext"
-  },
-  {
-    "algorithm": "RNA-FM / rna_fm_t12 RNA language model. Final layer 12 mean token pooling over real RNA tokens only; BOS/EOS excluded; long transcripts split into chunks and chunk means length-weighted.",
-    "category": "RNA",
-    "category_source": "manual",
-    "community_contribution": false,
-    "coverage_pct": 99.4,
-    "dataset_path": "data/embeddings/rnafm_full_transcript",
-    "description": "Mature-transcript representation generated from the final layer of RNA-FM using length-weighted token pooling.",
-    "id": "rnafm_full_transcript",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/rnafm_full_transcript/metadata.json",
-    "modality": "RNA language model / mature mRNA transcript",
-    "n_dims": 640,
-    "n_genes": 19976,
-    "name": "rnafm_full_transcript",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/rnafm_full_transcript",
-    "qc_status": "processed_qc_passed",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://github.com/ml4bio/RNA-FM"
-  },
-  {
-    "algorithm": "scFoundation / xTrimoGene (checkpoint: models.ckpt). fixed checkpoint embedding table; no cell/context pooling.",
-    "category": "Single-cell",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 94.85,
-    "dataset_path": "data/embeddings/scfoundation_gene_pos",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "scfoundation_gene_pos",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scfoundation_gene_pos/metadata.json",
-    "modality": "single_cell_expression_foundation_model / learned gene identity-position embedding",
-    "n_dims": 768,
-    "n_genes": 19062,
-    "name": "scFoundation-gene-pos",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scfoundation_gene_pos",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://github.com/biomap-research/scFoundation"
-  },
-  {
-    "algorithm": "scGPT whole-human input gene-token embedding layer. none; static learned token embedding weight extracted directly from checkpoint.",
-    "category": "Single-cell",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 95.93,
-    "dataset_path": "data/embeddings/scgpt_whole_human",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "scgpt_whole_human",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scgpt_whole_human/metadata.json",
-    "modality": "single_cell_expression_foundation_model",
-    "n_dims": 512,
-    "n_genes": 19280,
-    "name": "scGPT-whole-human",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scgpt_whole_human",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://github.com/bowang-lab/scGPT"
-  },
-  {
-    "algorithm": "scGPT pan-cancer checkpoint (checkpoint: scGPT pan-cancer gene-token embedding). none; static learned token embedding weight extracted directly from checkpoint.",
-    "category": "Single-cell",
-    "category_source": "manual",
-    "community_contribution": false,
-    "coverage_pct": 94.89,
-    "dataset_path": "data/embeddings/scgpt_pancancer",
-    "description": "Gene-token representation derived from a pan-cancer scGPT single-cell transcriptomic model.",
-    "id": "scgpt_pancancer",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scgpt_pancancer/metadata.json",
-    "modality": "single-cell transcriptomic foundation model / pan-cancer gene-token embedding",
-    "n_dims": 512,
-    "n_genes": 19071,
-    "name": "scgpt_pancancer",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scgpt_pancancer",
-    "qc_status": "processed_qc_passed",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://github.com/bowang-lab/scGPT"
-  },
-  {
-    "algorithm": "scPRINT (checkpoint: medium-v1.5.ckpt). fixed checkpoint embedding table; no cell/context pooling.",
-    "category": "Single-cell",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 99.49,
-    "dataset_path": "data/embeddings/scprint_medium_v1_5_gene",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "scprint_medium_v1_5_gene",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scprint_medium_v1_5_gene/metadata.json",
-    "modality": "single_cell_expression_foundation_model / learned gene-token embedding",
-    "n_dims": 256,
-    "n_genes": 19995,
-    "name": "scPRINT-medium-v1.5-gene",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/scprint_medium_v1_5_gene",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://huggingface.co/jkobject/scPRINT"
-  },
-  {
-    "algorithm": "SpliceAI splice-site prediction model. aggregate canonical splice-site profiles to one fixed-size vector per gene.",
-    "category": "Regulatory",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 100.0,
-    "dataset_path": "data/embeddings/spliceai_canonical_splice_site_profile",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "spliceai_canonical_splice_site_profile",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/spliceai_canonical_splice_site_profile/metadata.json",
-    "modality": "splicing sequence model / canonical transcript splice-site profiles",
-    "n_dims": 606,
-    "n_genes": 20097,
-    "name": "SpliceAI-canonical-splice-site-profile-zscore",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/spliceai_canonical_splice_site_profile",
-    "qc_status": "processed_qc_passed",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://github.com/Illumina/SpliceAI"
-  },
-  {
-    "algorithm": "Universal Cell Embeddings (UCE) (checkpoint: 33l_8ep_1024t_1280.torch). UCE gene encoder projection: Linear(5120->1280), ReLU, LayerNorm.",
-    "category": "Single-cell",
-    "category_source": "heuristic",
-    "community_contribution": false,
-    "coverage_pct": 94.95,
-    "dataset_path": "data/embeddings/uce_33l",
-    "description": "Core embedding package from embeddingsSSRF/gene_embeddings.",
-    "id": "uce_33l",
-    "license": "Original source / license terms",
-    "metadata_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/blob/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/uce_33l/metadata.json",
-    "modality": "single_cell_expression_foundation_model / protein-informed gene-token embedding",
-    "n_dims": 1280,
-    "n_genes": 19082,
-    "name": "UCE-33L",
-    "package_url": "https://huggingface.co/datasets/embeddingsSSRF/gene_embeddings/tree/326b257a20e73df337ece40bb82401e277851f5b/data/embeddings/uce_33l",
-    "qc_status": "legacy_core_package",
-    "source_dataset": "embeddingsSSRF/gene_embeddings",
-    "source_revision": "326b257a20e73df337ece40bb82401e277851f5b",
-    "zero_vectors": 0,
-    "license_url": "https://github.com/snap-stanford/UCE"
   }
 ]

release_summary.json

@@ -1,6 +1,14 @@
 {
-  "catalog_count": 69,
+  "catalog_count": 71,
   "community_catalog_count": 1,
   "community_registry_count": 1,
-  "latest_embedding_id": "hf_portal_root_upload_smoke_20260729_1328_v1"
+  "latest_embedding_id": "string_gnn",
+  "core_count": 70,
+  "community_count": 1,
+  "latest_embedding_ids": [
+    "string_wavegc",
+    "string_gnn"
+  ],
+  "core_dataset": "embeddingsSSRF/gene_embeddings",
+  "core_revision": "f703e997fa18de926e750ff9ec4ad452cbdd8184"
 }