InstaDeepAI

/

NTv3_tracks

Like 0

Follow

InstaDeep Ltd 451

Adds track_descriptions.csv, a table that describes every track the model predicts.

The model has one output channel per track, and each track is named by its id in the database it comes from, such as ENCSR056HPM or kai1. These ids are hard to read on their own. The table gives, for each id, the type of experiment and the sample it was measured on.

• One row per track id listed in track_metadata.json: 16,262 rows across the 15 species.
• Columns:
  • file_id: the track id.
  • assay: the type of experiment, for example total RNA-seq or ATAC-seq.
  • biosample_type and tissue: what was measured, for example a cell line and its name.
  • experiment_target: the protein or histone mark targeted, for ChIP-seq.
  • strand: the strand, for stranded assays.
  • species: the species.
  • dataset: the source database.

Example: ENCSR056HPM is total RNA-seq in the K562 cell line, from ENCODE.

The same file is added to InstaDeepAI/NTv3_tracks_1mb, which predicts the same tracks.

🤖 Generated with Claude Code