CoNCISE moodeng and moodeng-v2 dataset

The DTI dataset introduced in the CoNCISE paper. https://www.biorxiv.org/content/10.1101/2025.01.08.632039

We additionally add an extended dataset  that includes additional DTI information acquired from PLINDER.

moodeng-v1.tar.gz entries:

• train.csv (number of positives approximately 110000)
• test.csv
• val.csv

These files are comma separated CSV files. Each file has four fields:

Target Sequence: The protein sequence input

Label: A binary value. 1 indicates that the sequence binds with the ligand. O otherwise

SMILES: The ligand smiles

Remapped Entry: The representative sequence of the cluster the `Target Sequence` belongs to. Discard this field while doing DTI prediction

moodeng-v2-extended.tar.gz entries:

This is an updated dataset that with around 4 million entries (including the negative examples, negative:positive = 10:1; number of positives in train approximately 450000). The archive file contains two TAB separated CSV files: train.csv and test.csv

Following are their fields:

SMILES: same as above

Target Sequence: same as above

Label: Same as above

ID: A short name used to represent the `Target sequence`

cluster-assignment: The integer ID representing the cluster assignment.