LMDIPred predicts the occurrence of peptides representing linear motifs that bind to SH3, WW and PDZ domains, within user-provided amino-acid sequence(s), using four different prediction methods. A combined output table collects these independent predictions and lists the predicted peptides along with information about the method(s) that had predicted them.

Input: Users must provide fasta-formatted sequence(s) of one or more (upto 10) proteins or peptides containing more than 6 residues as input in the text box provided. Alternatively, ".txt" or ".fasta" files containing 1-10 protein/peptide sequences may also be uploaded as input.

Threshold: The threshold score for SVM prediction may be set by the user (default value 0.00, i.e, any contiguous stretch of amino acids with SVM prediction score above 0.00 will be predicted to bind to the chosen domain).

  • Choosing a higher threshold value makes the search more specific and reduces false positive hits, but some genuine hits might be missed.
  • Setting a lower threshold value, on the other hand, makes the search more sensitive and reduces false negatives, but may allow many spurious hits to be reported.


    1. Step 1: Please fill up all the required information


      2. Example Input:


    2. Step 2: Press SUBMIT button

      3. LMDIPred output represents a union of the prediction results from four independent methods in a tabular format, so that users can determine which peptide sequence was predicted by how many and which methods.

      • The output table lists the predicted linear peptide sequence and their starting and ending postions in the query, alongside the SVM prediction score/ PSSM score/ matching motif instance in LMPID/ matching regular expression, depending upon which prediction methods were chosen by the user.

      • A "---" sign in the output table means that the particular peptide was not predicted by that method.

      • Predictions by SVM and PSSM always represent 6-mer peptides for SH3 and WW domain, and 4-mer peptides for PDZ domain, because we have applied a sliding window approach with a window size of 6 and 4 respectively, for scanning of the input sequence by SVM or PSSM. However, peptides predicted by 'Motif Instance Matching' and 'Regular Expression Matching' may have variable length, according to the length of the matching Motif Instance or Regular Expression.


      Example Result:

    © 2015 Bose Institute, Kolkata. All rights reserved

    For queries and suggestions please contact Dr. Sudipto Saha (ssaha4@jcbose.ac.in, ssaha4@gmail.com)