Update data/embeddings/msa_transformer_depth1_uniprot_human/README.md
Browse files
data/embeddings/msa_transformer_depth1_uniprot_human/README.md
CHANGED
|
@@ -3,7 +3,6 @@
|
|
| 3 |
- Modality: Protein sequence / MSA Transformer depth-1 baseline
|
| 4 |
- Algorithm: FAIR-ESM MSA Transformer layer-12 mean-pooled protein representations using one UniProt sequence per one-row MSA
|
| 5 |
- Embedding source URL: https://github.com/facebookresearch/esm
|
| 6 |
-
- Original model/code URL: https://github.com/facebookresearch/esm
|
| 7 |
|
| 8 |
## Public files
|
| 9 |
|
|
@@ -50,8 +49,6 @@ Each input was a one-row MSA containing only the reference UniProt protein seque
|
|
| 50 |
model_checkpoint: esm_msa1b_t12_100M_UR50S
|
| 51 |
input_data_type: UniProt human reference proteome FASTA
|
| 52 |
raw_source: ~/data/raw_embeddings/msa_transformer_uniprot_human_v1_1/human_UP000005640_uniprot_proteome.fasta.gz
|
| 53 |
-
processing_script: ~/scripts/process_embeddings/process_msa_transformer_uniprot_single_sequence_msa_v1_1.py
|
| 54 |
-
environment: ~/envs/msa_transformer_embed
|
| 55 |
source_download_page: https://github.com/facebookresearch/esm
|
| 56 |
source_code_page: https://github.com/facebookresearch/esm
|
| 57 |
source_identifier_type: UniProt accession
|
|
@@ -65,9 +62,9 @@ Each input was a one-row MSA containing only the reference UniProt protein seque
|
|
| 65 |
|
| 66 |
### Source-processing decisions
|
| 67 |
|
| 68 |
-
The
|
| 69 |
|
| 70 |
-
The depth-1 design was intentional. A true homologous-MSA run would be biologically closer to the original purpose of MSA Transformer, but it would require generating or obtaining high-quality MSAs for roughly 19k human proteins.
|
| 71 |
|
| 72 |
Long proteins were split into non-overlapping windows of at most 1,022 amino acids. For each window, layer-12 residue representations were mean-pooled after removing special tokens. For proteins split across several windows, window embeddings were length-weighted and averaged into one final protein vector.
|
| 73 |
|
|
@@ -92,30 +89,7 @@ For this embedding:
|
|
| 92 |
|
| 93 |
This README describes the standardized gene-level version included in the public resource. The central requirement was that every retained row must align to one gene in master_gene_table.csv and that the matrix row order must match genes.tsv exactly.
|
| 94 |
|
| 95 |
-
This embedding is useful as a high-coverage protein-sequence baseline using the MSA Transformer architecture. In downstream analyses, compare it with ESM2 and other protein-sequence embeddings
|
| 96 |
-
|
| 97 |
-
### Final QC details
|
| 98 |
-
|
| 99 |
-
Status: processed_qc_passed
|
| 100 |
-
Master rows: 20,097
|
| 101 |
-
Source identifiers / source rows or entries: 20,652 FASTA indexed accessions; 19,472 mapped jobs attempted
|
| 102 |
-
Final mapped genes: 19,472
|
| 103 |
-
Master coverage: 96.8901%
|
| 104 |
-
Embedding shape: 19,472 x 768
|
| 105 |
-
FASTA indexed accessions: 20,652
|
| 106 |
-
Mapped jobs attempted: 19,472
|
| 107 |
-
Successful mapped genes: 19,472
|
| 108 |
-
NaN count: 0
|
| 109 |
-
Inf count: 0
|
| 110 |
-
Zero vectors: 0
|
| 111 |
-
Duplicated Ensembl IDs: 0
|
| 112 |
-
Failed sequence count: 0
|
| 113 |
-
Sequence length summary for successful genes:
|
| 114 |
-
count: 19,472
|
| 115 |
-
mean: 575.888044
|
| 116 |
-
min: 12
|
| 117 |
-
median: 428
|
| 118 |
-
max: 34,350
|
| 119 |
|
| 120 |
### Summary of final mapping:
|
| 121 |
|
|
|
|
| 3 |
- Modality: Protein sequence / MSA Transformer depth-1 baseline
|
| 4 |
- Algorithm: FAIR-ESM MSA Transformer layer-12 mean-pooled protein representations using one UniProt sequence per one-row MSA
|
| 5 |
- Embedding source URL: https://github.com/facebookresearch/esm
|
|
|
|
| 6 |
|
| 7 |
## Public files
|
| 8 |
|
|
|
|
| 49 |
model_checkpoint: esm_msa1b_t12_100M_UR50S
|
| 50 |
input_data_type: UniProt human reference proteome FASTA
|
| 51 |
raw_source: ~/data/raw_embeddings/msa_transformer_uniprot_human_v1_1/human_UP000005640_uniprot_proteome.fasta.gz
|
|
|
|
|
|
|
| 52 |
source_download_page: https://github.com/facebookresearch/esm
|
| 53 |
source_code_page: https://github.com/facebookresearch/esm
|
| 54 |
source_identifier_type: UniProt accession
|
|
|
|
| 62 |
|
| 63 |
### Source-processing decisions
|
| 64 |
|
| 65 |
+
The final input FASTA came from the UniProt FTP reference-proteome FASTA and was parsed with the final custom parser.
|
| 66 |
|
| 67 |
+
The depth-1 design was intentional. A true homologous-MSA run would be biologically closer to the original purpose of MSA Transformer, but it would require generating or obtaining high-quality MSAs for roughly 19k human proteins.
|
| 68 |
|
| 69 |
Long proteins were split into non-overlapping windows of at most 1,022 amino acids. For each window, layer-12 residue representations were mean-pooled after removing special tokens. For proteins split across several windows, window embeddings were length-weighted and averaged into one final protein vector.
|
| 70 |
|
|
|
|
| 89 |
|
| 90 |
This README describes the standardized gene-level version included in the public resource. The central requirement was that every retained row must align to one gene in master_gene_table.csv and that the matrix row order must match genes.tsv exactly.
|
| 91 |
|
| 92 |
+
This embedding is useful as a high-coverage protein-sequence baseline using the MSA Transformer architecture. In downstream analyses, compare it with ESM2 and other protein-sequence embeddings.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 93 |
|
| 94 |
### Summary of final mapping:
|
| 95 |
|