Extracting maximum likelihood estimates from a model_result

If you want to get the stats out-of a fitted model, use the evo.tabulate_stats() app.

We first fit a model.

[1]:

from cogent3.app import io, evo

loader = io.load_aligned(format="fasta", moltype="dna")
aln = loader("../data/primate_brca1.fasta")
model = evo.model("GN", tree="../data/primate_brca1.tree")
result = model(aln)

Create and apply tabulate_stats app

[2]:

tabulator = evo.tabulate_stats()
tabulated = tabulator(result)
tabulated

[2]:

3x tabular_result('global params': Table, 'edge params': Table, 'motif params': Table)

tabulated is a tabular_result instance which, like other result types, has dict like behaviour. It also contains key/value pairs for each model parameter type.

Edge parameters

These are all parameters that differ between edges. Since the current model is time-homogeneous (a single rate matrix), only the table only has entries for the branch scalar (denoted “length”).

[3]:

tabulated["edge params"]

[3]:
edge params
edge parent length
Galago root 0.1735
HowlerMon root 0.0450
Rhesus edge.3 0.0215
Orangutan edge.2 0.0078
Gorilla edge.1 0.0025
Human edge.0 0.0061
Chimpanzee edge.0 0.0028
edge.0 edge.1 0.0000
edge.1 edge.2 0.0033
edge.2 edge.3 0.0121
edge.3 root 0.0077

11 rows x 3 columns

NOTE: Unless the model is time-reversible, the lengths in that table are not ENS (Kaehler et al). As we used a non-stationary nucleotide model in this example, the length values are a scalar used to adjust the matrices during optimisation.

Global parameters

In this example, these are the elements of the rate matrix.

[4]:

tabulated["global params"]

[4]:
global params
A>C A>G A>T C>A C>G C>T G>A G>C G>T T>A
0.8700 3.6669 0.9111 1.5925 2.1264 6.0323 8.2178 1.2288 0.6294 1.2498
T>C
3.4136

1 rows x 11 columns

Motif parameters

In the current example, these are estimates of the nucleotide probabilities in the unobserved ancestor.

[5]:

tabulated["motif params"]

[5]:
motif params
A C G T
0.3756 0.1768 0.2078 0.2398

1 rows x 4 columns