1
$\begingroup$

The BUSCO _full_table.tsv output has scores per gene (an example bellow).

In the original publication, authors just mention a condition to satisfy a bitscore, but I could not find what the bitscore actually is. In the use guide they say

If found to be complete, whether single-copy or duplicated, the BUSCO matches have scored within the expected range of scores ...

but again, I don't understand what the score actually represents. I am even not sure if greater the better or visa reverse. Does anybody know what the score is?

# BUSCO version is: 4.0.2 
# The lineage dataset is: insecta_odb10 (Creation date: 2019-11-20, number of species: 75, number of BUSCOs: 1367)
# Busco id      Status  Sequence        Gene Start      Gene End        Score   Length
47at50557       Complete        scf1:20448-42284    20448   42284   4607.1  3554    https://www.orthodb.org/v10?query=47at50557     Immunoglobulin-like domain
547at50557      Complete        scf2:9780-23590    9780    23590   2743.8  1555    https://www.orthodb.org/v10?query=547at50557    protein unc-80 homolog
755at50557      Complete        scf3:9036-28083    9036    28083   3149.9  2206    https://www.orthodb.org/v10?query=755at50557    SRCR-like domain
Improve this question
$\endgroup$
6
  • $\begingroup$ Interesting question, its not immediately obvious I definitely agree $\endgroup$
    – M__ ♦
    Commented Mar 15, 2020 at 20:06
  • $\begingroup$ I don't think any of the guys from the Zdobnov lab are on this site yet, but do raise an issue on gitlab gitlab.com/ezlab/busco $\endgroup$
    – Chris_Rands
    Commented Mar 16, 2020 at 9:19
  • $\begingroup$ i recall that the bitscore cutoffs are HMMER3 thresholds that are tailored for each individual profile (i.e. model-specific), but since i never worked on BUSCO i might be wrong $\endgroup$
    – Chris_Rands
    Commented Mar 16, 2020 at 9:20
  • $\begingroup$ Thanks folks. If you are right about HMMER, then it is "A HMMER bit score is the log of the ratio of the sequence’s probability according to the profile (the homologyhypothesis) over the null model probability (the non-homology hypothesis)." $\endgroup$
    – Kamil S Jaron
    Commented Mar 16, 2020 at 9:38
  • $\begingroup$ I might as well invite Mathieu Seppey, I think it would be healthy to have him around :-). $\endgroup$
    – Kamil S Jaron
    Commented Mar 16, 2020 at 9:39

1 Answer 1

Reset to default
3
$\begingroup$

You are correct, it is the score of the candidate protein against the BUSCO profile hmm assessed by hmmer. If it passes a threshold, it is validated as being that BUSCO gene. The threshold is defined for each profile based on the score of the species used to produce the profile. It can be found in name_odb10/scores_cutoff. And yes, the higher the better.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Welcome, Mathieu! And thanks for the answer. $\endgroup$
    – Kamil S Jaron
    Commented Mar 16, 2020 at 14:52

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.