search

ikmb / MAGScoT

MAGScoT - a MAG scoring and refinement tool
GNU General Public License v3.0
27 stars 2 forks source link

MAGScoT give a bad MAG refinement result #5

Closed quliping closed 5 months ago

quliping commented 8 months ago

Hello, MAGScoT is an interesting software and is helpful for my metageome analysis. However, the refinement result of MAGScoT seems not good? I tried to analysis a sample using metaWrap and MAGScoT. I have three bin sets which were produced by maxbin2, metabat2, and vamb. Both MAGScoT (default parameters) and metaWRAP (bin_refinement module, completeness threshold 50% and contamination threshold 10%) produced five final MAGs, but results from MAGScoT seems have fewer species and very high contamination? Indeed MAGScoT takes less time and it's MAGs have higher completeness than metaWRAP, but such high contamination MAGs seems could not be used in article.

metaWRAP: image

MAGScoT: image

mruehlemann commented 8 months ago

Hi! Thank you for using MAGScoT. It looks like in this particular case of low numbers and high incompleteness of assembled genomes, the default scoring values in MAGScoT might not be perfectly fitting

MAGScoT uses three scoring parameters as described in the published article:

image

These parameters can be changed:

`

-a SCORE_A, --score_a=SCORE_A
    Scoring parameter a [default=1]

-b SCORE_B, --score_b=SCORE_B
    Scoring parameter b [default=0.5]

-c SCORE_C, --score_c=SCORE_C
    Scoring parameter c [default=0.5]

` metaWRAP has a much higher weight on duplicated SCG (b=5) (and lower weight on the multi SCG, c=0) if I remember correctly. Maybe you can try those values for MAGScoT and see if this improves your outcomes.

Best wishes, Malte