-
Notifications
You must be signed in to change notification settings - Fork 5
/
NEWS
84 lines (68 loc) · 4.2 KB
/
NEWS
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
CHANGES in VERSION 1.0.4
-------------------------
o Extended the vignette to explain the fitting of the expression priors better, especially for
Smart-seq data. For this, additional toy data for the tutorial was added to the package.
o Added the parameter "gene_length_prior" for the function "estimate.exp.prob.count.param" to
calculate the expression probabilities for Smart-seq data more accurately.
CHANGES IN VERSION 1.0.3
-------------------------
o Added the parameter reactionsPerKit to the Shiny Server.
CHANGES IN VERSION 1.0.2
-------------------------
o Added the Matrix package in order to run part of the functions with sparse matrices (however,
fitting itself is not implemented for sparse matrices yet, conversion still needed here)
o Added the number of 10X reactions per kit as an parameter called "reactionsPerKit" (before
always 6 rxns per kit)
o Added a function "print_optimalDesign_10X", which shows the optimal design in the end including
the number of required kits and flow cells
o Improved error message in FDR optimization if DE genes are specified wrong
CHANGES IN VERSION 1.0.1
-------------------------
o Corrected mistake in FDR optimization function (see Issue #4).
CHANGES IN VERSION 1.0.0
-------------------------
o Updated vignette reproduce-paper-plots to match with the updated plots of the publication.
CHANGES IN VERSION 0.1.3
-------------------------
o Added the option to calculate the power for unbalanced DE sets (using the variable
ssize.ratio.de) for all power calculation functions (not yet for the optimization).
The variable ssize.ratio.de represents the ratio of the group 1 sample size to
the group 0 sample size (control group). For this, the first argument of the
internal function power.de was changed from nSamples.group0 to nSamples (total sample
size).
o Added the option to use a different estimate for the number of independent SNPs per
locus using the variable indepSNPs (default is still 10). The parameter is
important for the Bonferroni multiple testing correction, as the total number of tests
to correct for is estimated by number of genes times number of independent SNPs
per locus.
o Added the option to set for the individual level expression threshold a percentage
threshold (expressed in x% of the cells for the individual and cell type) instead of the
absolute threshold (expressed with > x counts in sum for the individual and cell type)
using the parameter "cutoffVersion". Additionally, it is now also possible to set for
the population level expression threshold an absolute cutoff (expressed in > x of
individuals) instead of the percentage threshold (expressed in > x% of the individuals)
by setting the value of "perc.indiv.expr" to a number greater or equal than 1.
o Added a function to optimize the expression thresholds to maximize the power, called
"select.cutoffs". Please aware that by optimizing the power, potential a higher fraction
of false positives is to be expected.
o Improved shiny app layout (also visible in our website at scpower.helmholtz-muenchen.de)
o Included now a vignette "extension-complex-design" to show how to extend our method for
more complex experimental designs.
CHANGES IN VERSION 0.1.2
-------------------------
o Corrected bug for simulated eQTL power (significance parameter not processed correctly)
o Added the parameter "simThreshold" to define the maximal mean threshold for simulation
the eQTL power (default 4)
o Changed default value of parameter "useSimulatedPower" to FALSE for budget optimization
(this speeds the calculations significantly with no large effects on the optimal
parameter combinations). The functions starting with "optimize.constant.budget" are
affected.
o Removed the DESeq dependency by implementing the required methods within scPower (using
exactly the DESeq approach), as DESeq is deprecated on Bioconductor
CHANGES IN VERSION 0.1.1
-------------------------
o Added FDR as additional multiple testing option
o Increased power accuracy of eQTL genes with small mean values by simulation (option can
be turned off to speed up calculations)
o Option to skip power calculation of genes with very little expression probability to
speed up overall power calculation