forked from jwvdm/ebfret
-
Notifications
You must be signed in to change notification settings - Fork 2
/
CHANGELOG
62 lines (38 loc) · 2.41 KB
/
CHANGELOG
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Changelog for hFRET project
===========================
HEAD
----
* Added hstep_ml.m: A new method for doing hyperparameter updates. Instead of using the method of moments, this routine maximizes the total evidence over all traces with respect to the hyperparameters:
Grad_u Sum_n L_n = 0
* Added tracePlotter.m: A simple GUI that allows visualisation of VBEM output for all traces in a dataset.
* Added load_traces.m: A function to load in one or more datasets, remove photobleaching, as well as traces below a minimum length.
0.3 (still need to tag!)
------------------------
* get_ML_par: Parameter update for beta was based on wrong identity for mu_var. Previous update:
beta = 1 /(mu_var*lambda_mean)
New update:
beta = 1 / (mu_var * W * (v-2))
* synth_data_hFRET_subpop: Generates a set of synthetic traces with parameters drawn from specified priors.
* Added hFRET_subpop.m: New script to detect subpopulations.
1. Loads in traces from a number of '.dat' files (CSV format T rows, 2*N columns)
2. Removes photobleaching from traces by calling _photobleach_index_
3. Runs VBEM on each trace to get an initial set of output parameters for a number of states ranging from 1:K
4. Runs _get_ML_par_ to construct first guess for hyperparameters _u_ for number of states running from 1:K
5. For each configuration _s_, construct guesses for subpopulation hyperparameters by combining subblocks of guesses obtained in step 4
6. Run 1:R hFRET iterations for each configuration _s_
* Calculate max evidence parameters for each trace using VBEM
* Calculate new hyperparameters from VBEM parameters using method of moments
* Added hFRET_subpop_par.m and hFRET_subpop_cont.m: Parallelized version, and a script to resume a previously saved run.
* Added photobleach_index: Script to detect photobleaching point in a donor or acceptor signal.
1. Calculates running average moving forward through trace
2. Calculates running average running backward through trace
3. Take deviation between signal and forward/backward average
4. Look for correlated spike in deviation to identify bleaching point
* get_ML_par.m: Added optional argument LP for weighting of updates using evidence
0.2 (still need to tag!)
------------------------
* Created *util/* directory for supporting functions
* Moved scripts into separate dir. Removed all non-essential files
0.1
---
* Initial Import of Jonbron's hFRET code