Mirrors_Framework/deepmind-research
1
0
Fork 0
mirror of https://github.com/google-deepmind/deepmind-research.git synced 2026-05-09 21:07:49 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Fix documentation for the deletion_probability feature.

Browse Source 
PiperOrigin-RevId: 298798686
This commit is contained in:
Augustin Zidek
2020-03-04 09:45:08 +00:00
committed by Diego de Las Casas
parent 57e03c1bee
commit 985ca4dce7
1 changed files with 19 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
alphafold_casp13/README.md
+19 -7
View File
@@ -248,14 +248,26 @@ def sequence_to_onehot(sequence):
#### `deletion_probability`
The fraction of sequences that had a deletion (denoteby by a lowercase letter
in the A3M format) at this position. Example:
The fraction of sequences that had an insert state (denoted by a lowercase
letter in the A3M format) at this position. We used the following code to
compute it from the HHBlits MSA in the A3M format:
```
MSA = A c r k
A C r k
A C R k
deletion_probability = [[0], [1/3], [2/3], [1]]
```python
deletion_matrix = []
for msa_sequence in hhblits_a3m_sequences:
deletion_vec = []
deletion_count = 0
for j in msa_sequence:
if j.islower():
deletion_count += 1
else:
deletion_vec.append(deletion_count)
deletion_count = 0
deletion_matrix.append(deletion_vec)
deletion_matrix = np.array(deletion_matrix)
deletion_matrix[deletion_matrix != 0] = 1.0
deletion_probability = deletion_matrix.sum(axis=0) / len(deletion_matrix)
```
#### `gap_matrix`