Data to "Contrastive learning on protein embeddings enlightens midnight zone"

Published June 2022 | Version v1

Dataset Open

This is a backup of the data shared as part of the publication Contrastive learning on protein embeddings enlightens midnight zone and the associated github repo.

ProtTucker_ProtT5.pt holds model weights of ProtTucker
prottucker_training_embeddings.tar.gz holds input used to train ProtTucker (i.e., ProtT5 embeddings)
cath-domain-list.txt holds CATH annotation/labels used for training ProtTucker
pdb_seqres_161121.h5 holds pre-computed ProtTucker embeddings for PDB. Useful as lookup DB
scope_2.08_S100.h5 same as PDB_seqres but for SCOPe
sprot_161121.h5 same as PDB_seqres but for SwissProt
cath_v430_dom_seqs_S100_161121.h5 same as PDB_seqres but for CATH_v430

Files

Name	Size
cath-domain-list.txt md5:6df99c76b576d7fda8c47f0d43187ce1	37.1 MB	Preview Download
cath_v430_dom_seqs_S100_161121.h5 md5:2025cdd39c697de8b3cefffecfe35e09	299.9 MB	Download
pdb_seqres_161121.h5 md5:0dcb4e99bdae08229f64a2d187f65bdf	1.6 GB	Download
ProtTucker_ProtT5.pt md5:d558d1df229cf5ee0f864908ebc0500a	4.7 MB	Download
prottucker_training_embeddings.tar.gz md5:170a4ca81e258371c9afabbdfaa438e4	5.9 GB	Download
scope_2.08_S100.h5 md5:3cdbcd8362c757b7ea7af9ee993a0863	225.5 MB	Download
sprot_161121.h5 md5:0934353fd63182b66e862279c7985eb6	1.4 GB	Download