4
$\begingroup$

I am working on globular protein secondary structure assignments. The purpose of my project is to design an improved algorithm for secondary structure assignment.

I know two tools: DSSP and STRIDE for secondary structure prediction and assignment.

However, their accuracy [the Q3 (Q8) accuracy is defined as the percentage of residues for which the predicted secondary structures are correct] is around 53% and 70%, respectively, which isn't much if we compare them to modern algorithms like PALSSE, etc.

If so, why are DSSP and STRIDE considered reference algorithms, and why do people continue to use them?

Improve this question
$\endgroup$
6
  • $\begingroup$ This question is not specific enough. Please add more context to your post to describe a specific bioinformatics problem. Generic "What's the best tool" questions are better asked in discussion-type places like Reddit. $\endgroup$
    – gringer ♦
    Commented Apr 24 at 21:01
  • 1
    $\begingroup$ What does accuracy mean in this context? Comparison to human assignment? 53% for DSSP seems low to me. $\endgroup$
    – jgreener
    Commented Apr 29 at 9:36
  • 1
    $\begingroup$ @jgreener, The Q3 (Q8) accuracy is defined as the percentage of residues for which the predicted secondary structures are correct. $\endgroup$
    – user366312
    Commented Apr 29 at 12:35
  • 1
    $\begingroup$ But what is the source of ground truth? DSSP and STRIDE assign secondary structure from tertiary structure rather than predict it from sequence. I see the number 53% for DSSP in the PALSSE paper, but that is not for Q3 or Q8 accuracy. $\endgroup$
    – jgreener
    Commented Apr 29 at 13:30
  • 1
    $\begingroup$ @jgreener, then make some correction in the OP, and post an answer. $\endgroup$
    – user366312
    Commented Apr 29 at 15:13

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.