Instructions for Using MU-LOC Web Service

Step 1: Input sequences

Please input or upload your plant protein sequences in the FASTA format. If your input has only a single protein, the FASTA header line can be omitted. If you use both input box and upload button, we will use only the text from the input box.

For online submission, due to the computational cost of running PSI-BLAST, we limit at most 200 protein sequences per submission.

MU-LOC uses the first 22 amino acids to extract N-terminal features. Thus, MU-LOC does not accept sequences shorter than 22 amino acids.

MU-LOC expects each sequence has its unique name (unique fasta header line) since MU-LOC uses hash function to speed up the calculation of gene co-expression features. MU-LOC will trim each fasta header line to the first space if existed.

Step 2: Select estimated specificity

MU-LOC uses 4,500 non-mitochondrial and non-redundant plant proteins to estimate the false positive rate (also called specificity) of prediction.

For online submission, we provide 5 estimated specificity levels: 0.99, 0.95, 0.90, 0.85, and 0.80. MU-LOC uses 0.90 as the default specificity level.

For example, an estimated specificity of 0.90 represents that 10% of the 4,500 negative proteins are falsely predicted to be mitochondrial proteins.

Step 3: Choose a prediction program

MU-LOC provides two prediction algorithms, deep neural network (DNN) and support vector machine (SVM).

MU-LOC uses DNN by default. Please note that we use 10 trained DNNs and take the average prediction score from the 10 DNNs. Thus, the running time for DNN predictor would be longer than SVM.

Step 4 (Optional): Enter email address

MU-LOC relies on PSI-BLAST to generate position specific scoring matrix (PSSM) features. Also, MU-LOC searches over 20,000 Arabidopsis genes to calculate co-expression features. These processes might take a relatively long time for a large number of input sequences. To accommodate this, we provide an option for you to enter your email address. We will send you an email notification with the link to the prediction result when your job is finished.

Step 5: Job submit verification

We verify all the inputs and parameters you have provided. If everything is OK, we will submit your job automatically to the MU-LOC web service. Otherwise, we will mark the fields where we find errors. In this case, please check your input and resubmit.

Step 6: Job status and prediction results

If your job is successfully submitted, you can monitor the status of your job in Job Status section. We will print the job progress here. The Job Status section is automatically updated every 5 seconds. Also, you can manually refresh the webpage to check your job status.

Results will be printed as a table in the Prediction Results section when your job is successfully finished. We also provide a link to your prediction results in plain text format.

The output table/text file contains 3 columns, Sequence ID, Prediction Score, and Prediction Class. The prediction class is determined by the estimated specificity level you have chosen, with "mito" representing predicted mitochondrial protein, and "non-mito" as predicted non-mitochondrial protein.

You can also use the address ( to retrieve the status or prediction result of a previous job.