ChromBERT: Uncovering Chromatin State Motifs in the Human Genome using a BERT-based Approach

<ol> <li>Pretrain data and results for:&nbsp; <ul> <li>Promoter regions for all genes in 127 different cell lines in ROADMAP&nbsp;</li> <li>CRM (<em>cis</em>-regulatory module) regions longer than 2k bps in 127 different cell lines in ROADMAP</li> <li>Whole genome regions without continuous low signal state ("O") in 4-mer</li> </ul> </li> <li>Fine-tuning data and result for: <ul> <li>[Classification] Promoter regions of high-expressed genes (from RPKM&gt;10 to RPKM&gt;50) compared to not expressed genes (RPKM=0) or low-expressed genes (RPKM&gt;0) with the directory names:<br> <ul> <li>not_n_rpkm0 : RPKM=0 vs. RPKM&gt;0</li> <li>not_n_rpkm10 : RPKM=0 vs. RPKM&gt;10</li> <li>not_n_rpkm20 : RPKM=0 vs. RPKM&gt;20&nbsp;</li> <li>not_n_rpkm30 : RPKM=0 vs. RPKM&gt;30&nbsp;</li> <li>not_n_rpkm50 : RPKM=0 vs. RPKM&gt;50&nbsp;</li> <li>rpkm0_n_rpkm10 : RPKM&gt;0 vs. RPKM&gt;10&nbsp;</li> <li>rpkm0_n_rpkm20 : RPKM&gt;0 vs. RPKM&gt;20&nbsp;</li> <li>rpkm0_n_rpkm30 : RPKM&gt;0 vs. RPKM&gt;30&nbsp;</li> <li>rpkm0_n_rpkm50 : RPKM&gt;0 vs. RPKM&gt;50&nbsp;</li> <li>rpkm10_n_rpkm20 : RPKM&gt;10 vs. RPKM&gt;20&nbsp;</li> <li>rpkm10_n_rpkm30 : RPKM&gt;10 vs. RPKM&gt;30&nbsp;</li> <li>rpkm10_n_rpkm50 : RPKM&gt;10 vs. RPKM&gt;50&nbsp;</li> <li>rpkm20_n_rpkm30 : RPKM&gt;20 vs. RPKM&gt;30&nbsp;</li> <li>rpkm20_n_rpkm50 : RPKM&gt;20 vs. RPKM&gt;50&nbsp;</li> <li>rpkm30_n_rpkm50 : RPKM&gt;30 vs. RPKM&gt;50&nbsp;</li> </ul> </li> <li>[Regression] Quantitative gene expression prediction for promoter regions</li> <li>CRM&nbsp;regions longer than 2k bps compared to non-CRM regions</li> </ul> </li> </ol>