python UTR-LM/Scripts/UTRLM_downstream/predict_custom_utr.py \
--input-csv /path/to/your_utr.csv \
--output-csv /path/to/your_mrl_te_el_pred.csv \
--sequence-col utr \
--device cpu \
--predictor-modelfile UTR-LM/Model/Downstream/MRL/MJ3_seed1337_ESM2SISS_FS4.1.ep93.1e-2.dr5_unmod_1_utr_10folds_rl_LabelScalerFalse_LabelLog2False_AvgEmbFalse_BosEmbTrue_CNNlayer0_epoch300_nodes40_dropout30.5_finetuneTrue_huberlossTrue_lr0.01_fold0_epoch299.pt \
--esm2-modelfile UTR-LM/Model/Pretrained/ESM2SISS_FS4.1_fiveSpeciesCao_6layers_16heads_128embedsize_4096batchToks_lr1e-05_supervisedweight1.0_structureweight1.0_MLMLossMin_epoch93.pkl \
--output-col y_pred_mrl \
--te-predictor-modelfile UTR-LM/Model/Downstream/TE_EL/MJ4_seed1337_TE_ESM2SI_3.1.1e-2.H.dropout2_HEK_te_log_utr_seqlen100_AvgEmbFalse_BosEmbTrue_CNNlayer0_epoch300_patiences0_nodes40_dropout30.2_finetuneFalse_huberlossTrue_magicFalse_lr0.01_fold0_epoch28.pt \
--te-esm2-modelfile UTR-LM/Model/Pretrained/ESM2SI_3.1_fiveSpeciesCao_6layers_16heads_128embedsize_4096batchToks_MLMLossMin.pkl \
--el-predictor-modelfile UTR-LM/Model/Downstream/TE_EL/MJ4_seed1337_EL_ESM2SI_3.1.1e-2.H.dropout2_HEK_rnaseq_log_utr_seqlen100_AvgEmbFalse_BosEmbTrue_CNNlayer0_epoch300_patiences0_nodes40_dropout30.2_finetuneFalse_huberlossTrue_magicFalse_lr0.01_fold0_epoch217.pt \
--el-esm2-modelfile UTR-LM/Model/Pretrained/ESM2SI_3.1_fiveSpeciesCao_6layers_16heads_128embedsize_4096batchToks_MLMLossMin.pk --no-truncate
[predict_custom_utr.py](https://github.com/user-attachments/files/25875257/predict_custom_utr.py)
