Text this: A multi-modal co-attention model for accurate drug-target interaction prediction.