MM size: Macromolecular size

Number of chains in the model.

Stoi: Stoichiometry

Character representation of the assembly composition . For homooligomers it is just "A"+Macromol Size . For heterooligomers the stoichiometry is a combination of the number of chains coming from each target. A way to calculate this would be to look at the target models before concatenating, and report the number of chains in each one. The stoichiometry then would be calculated by sorting in descending order the list of sizes and attaching a letter of the alphabet before each size. For example: A2B2, or A3B2C1.

Symm: Symmetry

Character representation of the symmetry group.

Symm Size: Symmetry size

Number of chains involved in the symmetry.

Symm RMSD: Symmetry RMSD

The RMSD between chains using the symmetry axes.

Mdl conts

Number of contacts in the model.

Ref conts

Number of contacts in the reference structure.

Mdl clash

Number of clashes in the reference model.

ICS, Prec., Recall, IPS

The weighted sums of the corresponding measures (below) derived from all reproduced interfaces. The score for each chain-chain interface is weighted by the number of contacts in the corresponding interface in the target.

ICS

ICS is an F1-score representing the harmonic mean of the precision and recall. It is calculated only on the residues structurally resolved in the target. Formula:
ICS = F1 = 2*precision*recall/(precision+recall) * 100%.

Prec.

Precision of a multimeric prediction is the percentage of the correct interchain contacts among all interchain contacts in the model trimmed to the structurally resolved residues in the target. Formula:
Prec = TP/(TP+FP) * 100%,
where TP and FP are the numbers of correct and incorrect interchain contacts, correspondingly, in the predicted structure.

Recall

Recall of a multimeric prediction is the percentage of correctly reproduced native interchain contacts. The recall is calculated according to the formula:
Recall = TP/Nc *100%,
where TP is the number of correctly predicted interchain contacts,
Nc is the number of all interchain contacts in the target structure.

IPS

IPS is a Jaccard Coefficient showing similarity of the target and model interfaces. It is calculated on residues involved in interchain contacts and present in the target. Formula:
Jaccard Coefficient = (M11)/(M10+M10+M11),
where M01 is the number of the model interface residues that are not present in the target interface,
M10 is the number of the target interface residues that are not present in the model interface,
M11 is the number of interface residues that are present both in the model and the target interface.

QS_glob, QS_best

The QS scores (glob and best) are a Jaccard Coefficient showing similarity of target and model interface contacts. QS_glob accounts for all contacts in the target and the model (i.e. contacts covering any residues on the target sequence), while QS_best takes into consideration only residues structurally resolved in the target.

lDDT

The lDDT score evaluates the differences of the inter-atomic distances in model and target.

TM

The TM score evaluates overall similarity between the model and the target based on their optimal superposition.

RMSD

The global RMSD from the superposition of all chains based on the lowest RMSD chain matching.

DockQ(avg)

DockQ scores averaged with weights over all interfaces. DockQ [0,1] is a measure for assessing the accuracy of protein docking.