MU-LOC: A Machine-Learning Method for Predicting Mitochondrially Localized Proteins in Plants


MU-LOC is a machine learning method for protein mitochondrial targeting prediction in plants. MU-LOC is trained using both deep neural network (DNN) and support vector machine (SVM) on a comprehensive and high-quality training set.

Different types of features are extracted from protein sequences, including amino acid composition, sequence position specific scoring matrix, and gene co-expression information.

This site is an online web service of MU-LOC, with an Amazon Web Services (AWS) implementation, a standalone version of the software, and all the datasets we have collected and generated.


Using MU-LOC Web Service

MU-LOC web service is running normally.

Please input or upload your plant proteins below in the FASTA format, select estimated specificity, and MU-LOC predictor (DNN or SVM). For a single protein, the FASTA header (the ">" line) can be omitted. Please note that MU-LOC does not accept proteins shorter than 22 amino acids.

For online submission, due to the computational cost of running PSI-BLAST, we limit at most 200 protein sequences per submission. Also, since PSI-BLAST could take a while for some proteins, you can provide an email address where the prediction results will be sent to once the job is finished.

Here is a step-by-step instruction on how to use MU-LOC web service.

For a proteome wide prediction, please consider running MU-LOC on your local machine (download here) or Amazon Web Services (see instructions).

Using MU-LOC on Amazon Web Services

MU-LOC relies on some other tools and packages. Installing these dependencies could be complicated. As an alternative, we provide an implementation of MU-LOC on Amazon Web Services (AWS), where all the dependencies, environment variables, configuration files, etc., have been successfully set. You can run MU-LOC for proteome-wide predictions without worrying about installing any tools or setting paths for any data.

MU-LOC is available as an Amazon Machine Image (AMI: ami-12f37d6a) in the region of US West (Oregon). This page describes a detailed step-by-step instruction on how to launch this AMI and use MU-LOC as a cloud service.


Ning Zhang, R. Shyama Prasad Rao, Fernanda Salvato, Jesper F. Havelund, Ian Max Møller, Jay J. Thelen, and Dong Xu. 2018. MU-LOC: A Machine-Learning Method for Predicting Mitochondrially Localized Proteins in Plants. Submitted.


If you have any questions or comments, feel free to contact Ning Zhang (