Dataset Preview
Duplicate
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed
Error code:   DatasetGenerationError
Exception:    CastError
Message:      Couldn't cast
embedding: string
species: string
shape: list<item: int64>
  child 0, item: int64
dtype: string
gene_identifier: string
row_order: string
origin: struct<database: string, version: string, representation: string, portal: string, canonical_filename (... 197 chars omitted)
  child 0, database: string
  child 1, version: string
  child 2, representation: string
  child 3, portal: string
  child 4, canonical_filename: string
  child 5, canonical_sha256: string
  child 6, canonical_pipeline: string
  child 7, canonical_input_shape_models_genes: list<item: int64>
      child 0, item: int64
  child 8, local_prepared_input_sha256: string
  child 9, license: string
  child 10, provenance_note: string
preprocessing: struct<filtering: string, genes_removed_from_18531: int64, cell_lines_retained: int64, cell_lines_re (... 159 chars omitted)
  child 0, filtering: string
  child 1, genes_removed_from_18531: int64
  child 2, cell_lines_retained: int64
  child 3, cell_lines_removed: int64
  child 4, imputation: string
  child 5, imputed_genes: int64
  child 6, imputed_values: int64
  child 7, normalization: string
  child 8, pooling: string
  child 9, row_order: string
  child 10, column_order: string
identifier_mapping: struct<authority: string, url: string, sha256: string, snapshot_retrieved_utc: string, latest_record (... 520 chars omitted)
  child 0, authority: string
  child 1, url: string
  child 2, sha256: string
  child 3, snapshot_retrieved_utc: string
  child 4, latest_record_date_m
...
icense: string
  child 7, license_url: string
  child 8, collision_policy: string
  child 9, case_policy: string
  child 10, row_mapping: string
  child 11, method: string
  child 12, statistics: struct<rows: int64, resolved_to_hgnc: int64, unmapped: int64, ambiguous: int64, duplicate_canonical_ (... 230 chars omitted)
      child 0, rows: int64
      child 1, resolved_to_hgnc: int64
      child 2, unmapped: int64
      child 3, ambiguous: int64
      child 4, duplicate_canonical_symbols: int64
      child 5, rows_in_duplicate_symbol_groups: int64
      child 6, renamed_identifiers: int64
      child 7, exported_identifiers_that_are_current_hgnc_symbols: int64
      child 8, identifier_policies: struct<canonical_symbol: int64, original_duplicate_symbol: int64>
          child 0, canonical_symbol: int64
          child 1, original_duplicate_symbol: int64
files: struct<embeddings.npy: struct<sha256: string, bytes: int64>, genes.json: struct<sha256: string, byte (... 116 chars omitted)
  child 0, embeddings.npy: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 1, genes.json: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 2, models.json: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 3, row_mapping.csv: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
verification: string
to
{'text': Value('string')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1848, in _prepare_split_single
                  writer.write_table(table)
                  ~~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 783, in write_table
                  self.write_rows_on_file()  # in case there are buffered rows to write first
                  ~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 663, in write_rows_on_file
                  self._write_table(table)
                  ~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 795, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              embedding: string
              species: string
              shape: list<item: int64>
                child 0, item: int64
              dtype: string
              gene_identifier: string
              row_order: string
              origin: struct<database: string, version: string, representation: string, portal: string, canonical_filename (... 197 chars omitted)
                child 0, database: string
                child 1, version: string
                child 2, representation: string
                child 3, portal: string
                child 4, canonical_filename: string
                child 5, canonical_sha256: string
                child 6, canonical_pipeline: string
                child 7, canonical_input_shape_models_genes: list<item: int64>
                    child 0, item: int64
                child 8, local_prepared_input_sha256: string
                child 9, license: string
                child 10, provenance_note: string
              preprocessing: struct<filtering: string, genes_removed_from_18531: int64, cell_lines_retained: int64, cell_lines_re (... 159 chars omitted)
                child 0, filtering: string
                child 1, genes_removed_from_18531: int64
                child 2, cell_lines_retained: int64
                child 3, cell_lines_removed: int64
                child 4, imputation: string
                child 5, imputed_genes: int64
                child 6, imputed_values: int64
                child 7, normalization: string
                child 8, pooling: string
                child 9, row_order: string
                child 10, column_order: string
              identifier_mapping: struct<authority: string, url: string, sha256: string, snapshot_retrieved_utc: string, latest_record (... 520 chars omitted)
                child 0, authority: string
                child 1, url: string
                child 2, sha256: string
                child 3, snapshot_retrieved_utc: string
                child 4, latest_record_date_m
              ...
              icense: string
                child 7, license_url: string
                child 8, collision_policy: string
                child 9, case_policy: string
                child 10, row_mapping: string
                child 11, method: string
                child 12, statistics: struct<rows: int64, resolved_to_hgnc: int64, unmapped: int64, ambiguous: int64, duplicate_canonical_ (... 230 chars omitted)
                    child 0, rows: int64
                    child 1, resolved_to_hgnc: int64
                    child 2, unmapped: int64
                    child 3, ambiguous: int64
                    child 4, duplicate_canonical_symbols: int64
                    child 5, rows_in_duplicate_symbol_groups: int64
                    child 6, renamed_identifiers: int64
                    child 7, exported_identifiers_that_are_current_hgnc_symbols: int64
                    child 8, identifier_policies: struct<canonical_symbol: int64, original_duplicate_symbol: int64>
                        child 0, canonical_symbol: int64
                        child 1, original_duplicate_symbol: int64
              files: struct<embeddings.npy: struct<sha256: string, bytes: int64>, genes.json: struct<sha256: string, byte (... 116 chars omitted)
                child 0, embeddings.npy: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 1, genes.json: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 2, models.json: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 3, row_mapping.csv: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
              verification: string
              to
              {'text': Value('string')}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1850, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
                  ...<4 lines>...
                  )
              datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
              
              All the data files must have the same columns, but at some point there are 11 new columns ({'gene_identifier', 'preprocessing', 'row_order', 'verification', 'embedding', 'dtype', 'species', 'origin', 'identifier_mapping', 'files', 'shape'}) and 1 missing columns ({'text'}).
              
              This happened while the json dataset builder was generating data using
              
              hf://datasets/weililab/perturbation-embeddings/crispr_score_mean20/models.json (at revision 379bd319f295aafdfb98bbeda534fba92baa9085), ['hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_mean20/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_mean20/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_mean20/models.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_strict/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_strict/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_strict/models.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/esm2/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/esm2/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/gears/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/gears/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/genept/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/genept/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_net/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_net/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_net/proteins.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_prot/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_prot/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_prot/proteins.json'], ['hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_mean20/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_mean20/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_mean20/models.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_strict/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_strict/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/crispr_score_strict/models.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/esm2/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/esm2/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/gears/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/gears/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/genept/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/genept/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_net/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_net/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_net/proteins.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_prot/genes.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_prot/metadata.json', 'hf://datasets/weililab/perturbation-embeddings@379bd319f295aafdfb98bbeda534fba92baa9085/string_prot/proteins.json']
              
              Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1868, in _prepare_split_single
                  num_examples, num_bytes = writer.finalize()
                                            ~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 803, in finalize
                  self.write_rows_on_file()
                  ~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 663, in write_rows_on_file
                  self._write_table(table)
                  ~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 795, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              embedding: string
              species: string
              shape: list<item: int64>
                child 0, item: int64
              dtype: string
              gene_identifier: string
              row_order: string
              origin: struct<database: string, version: string, representation: string, portal: string, canonical_filename (... 197 chars omitted)
                child 0, database: string
                child 1, version: string
                child 2, representation: string
                child 3, portal: string
                child 4, canonical_filename: string
                child 5, canonical_sha256: string
                child 6, canonical_pipeline: string
                child 7, canonical_input_shape_models_genes: list<item: int64>
                    child 0, item: int64
                child 8, local_prepared_input_sha256: string
                child 9, license: string
                child 10, provenance_note: string
              preprocessing: struct<filtering: string, genes_removed_from_18531: int64, cell_lines_retained: int64, cell_lines_re (... 159 chars omitted)
                child 0, filtering: string
                child 1, genes_removed_from_18531: int64
                child 2, cell_lines_retained: int64
                child 3, cell_lines_removed: int64
                child 4, imputation: string
                child 5, imputed_genes: int64
                child 6, imputed_values: int64
                child 7, normalization: string
                child 8, pooling: string
                child 9, row_order: string
                child 10, column_order: string
              identifier_mapping: struct<authority: string, url: string, sha256: string, snapshot_retrieved_utc: string, latest_record (... 520 chars omitted)
                child 0, authority: string
                child 1, url: string
                child 2, sha256: string
                child 3, snapshot_retrieved_utc: string
                child 4, latest_record_date_m
              ...
              icense: string
                child 7, license_url: string
                child 8, collision_policy: string
                child 9, case_policy: string
                child 10, row_mapping: string
                child 11, method: string
                child 12, statistics: struct<rows: int64, resolved_to_hgnc: int64, unmapped: int64, ambiguous: int64, duplicate_canonical_ (... 230 chars omitted)
                    child 0, rows: int64
                    child 1, resolved_to_hgnc: int64
                    child 2, unmapped: int64
                    child 3, ambiguous: int64
                    child 4, duplicate_canonical_symbols: int64
                    child 5, rows_in_duplicate_symbol_groups: int64
                    child 6, renamed_identifiers: int64
                    child 7, exported_identifiers_that_are_current_hgnc_symbols: int64
                    child 8, identifier_policies: struct<canonical_symbol: int64, original_duplicate_symbol: int64>
                        child 0, canonical_symbol: int64
                        child 1, original_duplicate_symbol: int64
              files: struct<embeddings.npy: struct<sha256: string, bytes: int64>, genes.json: struct<sha256: string, byte (... 116 chars omitted)
                child 0, embeddings.npy: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 1, genes.json: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 2, models.json: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 3, row_mapping.csv: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
              verification: string
              to
              {'text': Value('string')}
              because column names don't match
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      builder, max_dataset_size_bytes=max_dataset_size_bytes
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  ):
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1879, in _prepare_split_single
                  raise DatasetGenerationError("An error occurred while generating the dataset") from e
              datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

text
string
A1BG
A1CF
A2M
A2ML1
A3GALT2
A4GALT
A4GNT
AAAS
AACS
AADAC
AADACL2
AADACL3
AADACL4
AADAT
AAGAB
AAK1
AAMDC
AAMP
AANAT
AAR2
AARD
AARS1
AARS2
AARSD1
AASDH
AASDHPPT
AASS
AATF
LMTK1
ABAT
ABCA1
ABCA10
ABCA12
ABCA13
ABCA2
ABCA3
ABCA4
ABCA5
ABCA6
ABCA7
ABCA8
ABCA9
ABCB1
ABCB10
ABCB11
ABCB4
ABCB5
ABCB6
ABCB7
ABCB8
ABCB9
ABCC1
ABCC10
ABCC11
ABCC12
ABCC2
ABCC3
ABCC4
ABCC5
ABCC8
ABCC9
ABCD1
ABCD2
ABCD3
ABCD4
ABCE1
ABCF1
ABCF3
ABCG1
ABCG2
ABCG4
ABCG5
ABCG8
ABHD1
ABHD10
ABHD11
ABHD12
ABHD12B
ABHD13
ABHD14A
ABHD14B
ABHD15
ABHD16A
ABHD16B
ABHD17A
ABHD17B
ABHD17C
ABHD18
ABHD2
ABHD3
ABHD4
ABHD5
ABHD6
ABHD8
ABI1
ABI2
ABI3
ABI3BP
ABITRAM
ABL1
End of preview.

Gene embeddings for perturbation modeling

This repository distributes seven gene/protein representation tables in a common NumPy/JSON format for use with pertTF and other perturbation models. The STRING, ESM2, GenePT and GEARS tables are embedding features. The CRISPR score tables contain measured GeneEffect profiles across cell lines, used as gene features. None of these tables contains pertTF-generated predictions. The original embedding models were not trained by this release.

Tables

Directory Representation Rows Dimensions Published dtype
esm2 ESM2 protein embeddings distributed with UCE 19,516 5,120 float32
genept GenePT gene/protein text embeddings, model-3 release 19,134 3,072 float32
gears Learned perturbation embeddings from a legacy GEARS checkpoint 9,853 32 float32
string_net STRING protein network embeddings 19,699 512 float32
string_prot STRING ProtT5 protein sequence embeddings 19,699 1,024 float32
crispr_score_strict Complete-case CRISPR GeneEffect profiles 17,087 1,208 float32
crispr_score_mean20 CRISPR GeneEffect profiles, <=20% missingness and gene-mean imputation 17,787 1,208 float32

Each directory contains:

  • embeddings.npy: a dense matrix with shape (n_genes, embedding_dim).
  • genes.json: an ordered JSON list; genes[i] labels row i in the matrix.
  • metadata.json: source identifiers, processing details, file hashes, and conversion verification results.

ESM2 and GenePT are restricted to gene symbols in the first column of their respective human_prot_names.csv files. Only listed genes present in each embedding table are retained, preserving original gene identifiers and row order. Gene-list hashes, matched counts, and unmatched counts are recorded in each table's metadata. GEARS is unchanged. These three original tables had no normalization, identifier conversion, concatenation, or synthetic control (WT) row added. The four additions use the identifier policy described below.

This corrects the initial release, which included the complete unfiltered GenePT mapping. The earlier revision 5db008c45c77caaf5ce9e7c9783c89ce1d8a2f25 remains available for reproducing that release.

The ESM2 and GEARS inputs were float32. The GenePT input was float64, but its values were exactly representable in float32: the maximum absolute conversion error was zero for all three tables. Every exported row was checked against its corresponding source vector after reloading the output files, and the saved gene ordering was verified.

Loading

These are ordinary files hosted in an HF dataset repository; the datasets package is not required.

import json
import numpy as np
from huggingface_hub import HfApi, hf_hub_download

repo_id = "weililab/perturbation-embeddings"
embedding = "string_net"  # any directory in the table above
# Resolve once so the matrix and gene list come from the same snapshot.
# Save this commit SHA with your experiment, or supply a previously saved SHA.
revision = HfApi().dataset_info(repo_id).sha

matrix_path = hf_hub_download(
    repo_id, f"{embedding}/embeddings.npy",
    repo_type="dataset", revision=revision,
)
genes_path = hf_hub_download(
    repo_id, f"{embedding}/genes.json",
    repo_type="dataset", revision=revision,
)
matrix = np.load(matrix_path, mmap_mode="r", allow_pickle=False)
with open(genes_path) as handle:
    genes = json.load(handle)
assert matrix.shape[0] == len(genes)

Gene coverage differs between tables. Align by gene identifier rather than row position when comparing or concatenating representations. A trained model's input representation cannot generally be replaced with another embedding choice without retraining.

Sources, attribution, and licenses

Licensing is recorded per source; no single license is asserted for all tables.

ESM2 / UCE

  • Source record: Yusuf Roohani (2023), Universal Cell Embedding Model Files, Figshare, version 5: https://doi.org/10.6084/m9.figshare.24320806.v5.
  • The record distributes protein_embeddings.tar.gz and specifies CC BY 4.0.
  • The local input was Homo_sapiens.GRCh38.gene_symbol_to_embedding_ESM2.pt, associated with that UCE download. This release did not byte-compare the individual file against the original archive.
  • Credit the UCE authors and the ESM2 authors when using these embeddings. ESM2 reference: Lin et al. (2023), Evolutionary-scale prediction of atomic-level protein structure with a language model, https://doi.org/10.1126/science.ade2574.

GenePT

GEARS

  • Extracted pert_emb.weight from the legacy checkpoint gears_gwps-Mar25-03-38/best_model/model.pt, using the ordered pert_gene_list from its accompanying pert_gene_list.pkl.
  • Training provenance is unknown. The training dataset, split, and exact training configuration were not reconstructed for this release.
  • The checkpoint's license is unknown; the CC BY 4.0 licenses of the other sources are not attributed to this checkpoint.
  • GEARS project: https://github.com/snap-stanford/GEARS.

STRING network (string_net)

  • Source: STRING downloads. Exact source release, download URL and file hash are in metadata.json.
  • Cross-species-aligned SPACE protein network embeddings: 19,699 proteins, each represented by a 512-dimensional vector.
  • All source rows are retained. Original float16 values are exactly preserved in float32. proteins.json records the original protein IDs in row order.
  • The STRING download page licenses this file under CC BY 4.0.
  • Please cite SPACE (https://doi.org/10.1093/bioinformatics/btaf496) and STRING (https://doi.org/10.1093/nar/gkae1113).

STRING protein sequence (string_prot)

  • Source: STRING downloads. Exact source release, download URL and file hash are in metadata.json.
  • ProtT5 protein sequence embeddings: 19,699 proteins, each represented by a 1,024-dimensional vector. These complement the network-derived features.
  • All source rows are retained. Original float16 values are exactly preserved in float32. proteins.json records the original protein IDs in row order. Network and sequence rows have different orders; align by identifier.
  • The STRING download page licenses this file under CC BY 4.0.
  • Please cite SPACE (https://doi.org/10.1093/bioinformatics/btaf496) and STRING (https://doi.org/10.1093/nar/gkae1113).

CRISPR GeneEffect, complete-case (crispr_score_strict)

  • Source: the canonical DepMap GeneEffect table with 18,531 genes and 1,208 cell lines. Exact release and canonical input identity are in metadata.json.
  • Each gene is represented by its measured GeneEffect scores across cell lines, rather than a learned lower-dimensional vector. models.json identifies the cell lines in feature-column order.
  • Retain genes finite in every cell line in both aligned GeneEffect and GeneDependency tables. The missingness masks were identical.
  • Remove 1,444 genes, yielding 17,087 genes; retain all 1,208 cell lines. No imputation is performed.

CRISPR GeneEffect, mean-imputed (crispr_score_mean20)

  • Source: the same canonical DepMap GeneEffect table used for crispr_score_strict. Exact release and canonical input identity are in metadata.json.
  • Each gene is represented by its measured GeneEffect profile across the same 1,208 cell lines, with missing entries filled as described below. models.json identifies the cell lines in feature-column order.
  • Remove genes with >20% missing GeneEffect values. For retained genes, fill missing values with that gene's observed mean across cell lines.
  • Remove 744 genes, yielding 17,787 genes; impute 79,974 values across 700 additional genes. All original complete-case gene vectors remain unchanged.

For both CRISPR tables, no additional centering, scaling, clipping or normalization is applied. Release identity is recorded from the local canonical pipeline; the canonical input was not byte-compared with the original portal download. The exact release's licensing was not independently verified and is recorded accordingly in metadata; licenses from the other sources are not assigned to these tables.

Processing is source-specific: human-gene filtering of the original tables, format conversion, float32 storage, HGNC identifier mapping for the additions, and the explicitly described CRISPR filtering/imputation. Details and source identities are recorded in each table's metadata.

Naming conventions and identifier mapping

Directory names identify the representation: string_net, string_prot, crispr_score_strict, and crispr_score_mean20. Specific data origins and versions are recorded in each table's metadata.json.

These four additions use current HGNC symbols where the mapping is unambiguous and one-to-one. Every original row and vector is retained. All members of a duplicate-symbol group keep their original identifiers, as do unresolved or ambiguous rows. Consequently, genes.json contains unique lookup keys but is not exclusively a list of current HGNC symbols. No vectors are averaged or merged.

Each addition includes row_mapping.csv, aligned with the matrix rows, recording original identifiers, canonical symbols, HGNC IDs, mapping evidence, and the identifier policy.

The tables retain full source coverage rather than a comparison-specific intersection. Source releases, download locations, exact input hashes, the HGNC snapshot identity, mapping methods and counts are in metadata.json. The three previously published tables retain their existing identifiers.

HGNC mapping data are CC0: https://www.genenames.org/about/license/. Please credit HGNC (RRID:SCR_002827).

Downloads last month
271