-
Notifications
You must be signed in to change notification settings - Fork 1
/
CHANGES.txt
257 lines (171 loc) · 7.19 KB
/
CHANGES.txt
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
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
SPAN Changelog
===============
Here you can see the full list of changes between each release.
Version 2.0
----------------
Released on November 11, 2024
- Improved performance in various signal-to-noise conditions with data-estimated constraints of signal-to-noise ratio and minimal noise state threshold
- Improved constrained model fitting states adjustment
- Fixed aggressive signal upscaling in case of exceeding control library sizes
- Use global genome signal and noise coverage estimation for boundaries cut
- Support blacklisted regions `--blacklist`
- Support `--deep-analysis` additional analysis of peak calling producing important statistics
- Extended logging in debug mode `--debug`
- Configure multiple hypothesis adjustment method with `--multiple`
- Sensitivity and gap are computed automatically based on experimental data, but can be configured manually with `--sensitivity` and `--gap`
Version 1.6
----------------
Released on Apr 30, 2024
- Faster SPAN model convergence by relative likelihood estimation
- Improved initialization to make SPAN more robust in low signal-to-noise ratio case
- Don't process contigs when coverage distribution is different from genome-wide
- Introduced `--bg-sensitivity` to configure sensitivity of peak backgrounds
- Dropped `gap` as obsolete
- Peaks from model procedure significantly improved tuning flexibility
Version 1.5
----------------
Released on Jan 25, 2024
- Specify log file location with `--log` command line option
- Don't keep cache files until `--keep-cache` parameter is provided
Version 1.4
----------------
Released on Nov 2, 2023
- Fixed problem with SPAN ignoring --threads command line argument
Version 1.3
----------------
Released on July 20, 2023
- Fixed problem with paired end coverage loading and too long paired fragments
Version 1.2
----------------
Released on Jun 27, 2023
- Improved peak calling for narrow marks histone marks
- Improved peaks visualization in JBR Genome Browser
- Fixed beta value estimation for control correction on chromosome filtered coverage
- Fixed problem with highly relaxed FDR settings
Version 1.1
----------------
Released on Mar 9th, 2023
- Bugfix update - fix problem with treatment and control scaling
- Upscale treatment and control library
- Slightly tweaked defaults - number of iterations and threshold till convergence
- Reverted compatibility with Java 8
Version 1.0
----------------
Released on Oct 12th, 2022
- Switched to new procedure for computing peaks from HMM:
* 1) Estimate HMM posterior error probabilities
* 2) Pick candidate bins with relaxed posterior error probability, e.g. sqrt(fdr).
* This mitigates the problem of wide marks peaks split on strong fdrs.
* 3) Using gap merge bins into candidate islands.
* 4) Assign p-value to each island using based on combined p-values for blocks of consequent enriched bins.
* Each block is assigned P as average posterior log error probability for bins in blocks.
* 50% top significant blocks scores are aggregated using length-weighted average as P for island.
* 5) Compute qvalues on islands p-values, filter by alpha.
- Improved treatment and control normalization, use beta minimizing absolute correlation between
corrected signal and control: |correlation(scaled-treatment - beta * scaled-control, scaled-control)|
- Improved multistart initial NB states guess - use different params for each model
- Safeguard against infinite failures as initial guess for NB, which effectively converts NB to Poisson
- Show progress while computing peaks from model
- Updated defaults values for SPAN peaks: fdr 0.05, gap 3
- Updated defaults for SPAN model: total iterations 20, log-likelihood threshold 1
- Reduced time required for peak calling
- Added advanced SPAN model viewer in JBR genome browser
- Bugfixes
Version 0.13
----------------
Released on Aug 12th, 2020
- Bugfixes in log formats
Version 0.12.0
----------------
Released on Apr 21th, 2020
- Fixed multistart procedure
- Switched to SL4J and Backlog logging
Version 0.11.0
-------------
Released on May 17th, 2019
- If the model file is present, Span can now call and tune peaks with no further options required
- If the options are nevertheless provided, they're checked against those stored in the model file
- Span recognizes explicit options `--fragment auto` and `--keep-dup false` (previously, the user had to
omit the options to achieve this result)
- Reduced statistical data model convergence time
- Changed model initial point
- Multistart for model fitting
- Safeguards against NB parameters vanishing for small signal-to-noise ratio or insufficient number of reads
Version 0.10.0
-------------
Released on March 22, 2019
- ATAC-Seq support - `--fragment 0` option works correctly in paired end reads case
- Single Cell ATAC-Seq supported - `--keep-dup` options allows duplicated alignments in same position
- SPAN model name can be configured explicitly by `--model`, default name is more user-friendly
- SPAN doesn't ignore contigs (alternative, random) chromosomes in peak calling anymore
Version 0.9.2
-------------
Released on Jan 16, 2019
- Fixed viktor library version
Version 0.9.1
-------------
Released on Dec 14, 2018
- Fixed a bug with SPAN reading a BAM file twice and creating two caches
Version 0.9.0
-------------
Released on Dec 7, 2018
- Paired-end BAM library support
- Estimate fragment size for single-end libraries using maximum cross-correlation approach
by Kharchenko et al., 2008 (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2597701/)
Version 0.8.0
-------------
Released on Nov 22, 2018
- SPAN is available on Galaxy, see: https://github.com/JetBrains-Research/galaxy-applications
- Ignore empty chromosomes while fitting models
- Lightweight models can be opened in JBR Genome Browser
- Improved support for custom chrom.sizes path
- Windows, macOS supported in Viktor (https://github.com/JetBrains-Research/viktor) enabling SS2, AVX instructions set
- Various bugfixes
Version 0.7.0
-------------
Released on ?
- Renamed to SPAN (Semi-supervised Peak Analyzer)
Version 0.6.0
-------------
Released on May 31th, 2018
- Save resulting model to tar file, which can be uploaded to JBR genome browser
- Various bugfixes
Version 0.5.1
-------------
Released on January 30th, 2018
- Use median p- and q-value for peaks
Version 0.5.0
-------------
Released on January 24th, 2018
- Use single model for all chromosomes
- Utilize concurrency
Version 0.4.0
-------------
Released on July 27th, 2017
- Fragment size support.
- Control support.
- Use gap to union close narrow peaks into a broad ones.
Version 0.3.0
-------------
Released on May 10th, 2017
- New CLI: bed and fdr are now obligatory arguments.
- Instead of 2bit reference we now use chrom.sizes.
Version 0.2.3
-------------
Released on February 10th, 2016
- Show meaningful error message on uncorrelated replicates.
Version 0.2.2
-------------
Released on January 19th, 2016
- Support BAM input file format.
Version 0.2.1
-------------
Released on January 11th, 2016
- Fixed an issue with difference asymmetry.
Version 0.2.0
-------------
Released on December 16th, 2015
- Fixed an issue with possible states switching in comparison analysis.
Version 0.0.1
-------------
Initial release, released on September 15th, 2015