Repository navigation
Expand file tree
/
Copy pathMANUAL
More file actions
117 lines (91 loc) · 5.32 KB
/
Copy pathMANUAL
File metadata and controls
117 lines (91 loc) · 5.32 KB
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
===============
RiboDiff Manual
===============
This document shows how to use RiboDiff to detect the protein translational efficiency
change from ribosome footprint profile (Ribo-Seq) and RNA-Seq data in different
experimental conditions.
To run RiboDiff, please provide two input files:
1) A comma delimited text file (CSV) that describes the experimental design. Here is an example:
Samples,Data_Type,Conditions
RbCtlR1,Ribo-Seq,Control
RbCtlR2,Ribo-Seq,Control
RbTrtR1,Ribo-Seq,DrugTreated
RbTrtR2,Ribo-Seq,DrugTreated
RnaCtlR1,RNA-Seq,Control
RnaCtlR2,RNA-Seq,Control
RnaCtlR3,RNA-Seq,Control
RnaTrtR1,RNA-Seq,DrugTreated
RnaTrtR2,RNA-Seq,DrugTreated
RnaTrtR3,RNA-Seq,DrugTreated
In this experimental design file, there are three columns organized underneath a header
line. The first column indicates the sample names. The second column indicates the data
type. It only accepts "Ribo-Seq" and "RNA-Seq" to represent the ribosome footprint and
RNA-Seq data, respectively. The last column indicates to which condition the sample
belongs. More than two condition keywords are not accepted here. Note the columns are
comma separated.
2) A text file containing the raw read count for each gene as follows:
Entry RbCtlR1 RbCtlR2 RbTrtR1 RbTrtR2 RnaCtlR1 RnaCtlR2 RnaCtlR3 RnaTrtR1 RnaTrtR2 RnaTrtR3
G0001 69 77 81 98 2914 2931 3048 2520 2566 2464
G0002 159 141 145 139 1285 1285 1246 1242 1246 1313
G0003 246 239 236 259 806 847 859 862 819 833
G0004 70 59 71 84 1413 1490 1511 1464 1499 1544
.....
.....
In this count file, the sample name in header line must agree with that in experimental design
file. The columns are tab-separated. Only including protein coding genes may give a better
global estimation.
Assuming you are at the RiboDiff package directory, the command to run RiboDiff is:
python ./scripts/TE.py -e <exp_outline.txt> -c <cnt_table.txt> -o <result.txt>
RiboDiff loads the input files and normalizes the raw count of each replicate sample
according to its sequencing library size. Then, it estimates the dispersion parameter
for each gene using a generalized linear model, assuming that read counts follow a negative
binomial distribution. Note, users can enable/disable the different dispersion estimation
for Ribo-Seq and RNA-Seq separately. Next, RiboDiff performs a statistical test based
on the H0 and H1 model fitting. Finally, multiple test correction is performed to generate
an adjusted P value for every gene.
RiboDiff also calculates the log2 fold change of translational efficiency for all genes.
In addition, RiboDiff can plot figures to visualize the data and the statistical test
results. The figures includes
1) A scatter plot of dispersion against the mean count across replicates of Ribo-Seq
and RNA-Seq.
2) A histogram of dispersion difference of all genes.
3) A scatter plot of log fold change of translational efficiency against the mean
count across replicates of Ribo-Seq.
4) A histogram of log fold change of translational efficiency with significant proportion
marked in red (TE down) and blue (TE up).
The dispersion used in 1) and 2) is the empirical dispersion. It is calculated as the
relationship of Var = mean + (mean x disp^2)
Here is the help information of running RiboDiff. Users can display it by typing
"python TE.py -h" in command line.
Usage: TE.py [options]
Options:
-h, --help show this help message and exit
REQUIRED:
-e EXPTOUTLINE Text file describing experiment Outline. Must follow
required format, please see the manual.
-c CNTFILE Text file containing the count data. Header line must be
consistent with information in experiment Outline.
-o OUTFILE Tab delimited text file containing the results.
OPTIONAL:
-d DISPDIFF Allow different dispersions for Ribo-seq and RNA-seq
count data. On: 1; Off: 0. [default: 1]
-s SUMCNTCUTOFF Set the sum of normalized read count as the threshold to
do the test. This option applies for both Ribo-seq and
RNA-seq data. [default: 10]
-i DISPINITIAL Set the initial dispersion to start the estimation.
[default: 0.01]
-m MULTITEST Method for multiple test correction. Options: BH
(Benjamini-Hochberg); Bonferroni. [default: BH]
-r RANKRESULT Rank the result table in ascending order by a specific
column. Adjusted p value: 1; TE change: 2; Gene id: 3;
Keep the order as in count file: 0. [default: 0]
-p PLOTS Make plots to show the data and results. Plots are in pdf
format. On: 1; Off: 0. [default: 0]
-q CUTOFFFDR Set the FDR cutoff for significant case to plot.
[default: 0.1]
Similar to many other RNA-Seq based tools, RiboDiff uses negative binomial distribution to
model the read count, which handles larger variation across samples than Poisson. However,
if the randomness of count data from certain types of samples are extremely large, limited
number of replicates cannot provide a good estimation on dispersion, which ends up with
less significant results. Also, we suggest the users to figure out the correlation between
replicates. Removing the outliers of replicates before running RiboDiff is recommended.