Open lixin4306ren opened 5 years ago
you can extract the expression matrix and cell-gene data from EMSet, and create an scGPS object
Thank you for your reply. I finished scGPS clustering analysis and compared the results with that from ascend. For the same sample, I found that ascend generated 3 clusters, while scGPS's optimal result generated 10 clusters. How to explain the difference and which result should I use? Thanks.
ascend results:
> nor2@clusterAnalysis$keyStats
Height Stability RandIndex ConsecutiveRI ClusterCount
1 0.025 0.200 1.0000000 1.0000000 9
2 0.05 0.200 1.0000000 1.0000000 9
3 0.075 0.200 1.0000000 1.0000000 9
4 0.1 0.200 1.0000000 1.0000000 9
5 0.125 0.200 1.0000000 1.0000000 9
6 0.15 0.200 1.0000000 1.0000000 9
7 0.175 0.200 1.0000000 1.0000000 9
8 0.2 0.200 1.0000000 1.0000000 9
9 0.225 0.025 0.8231890 0.8231890 8
10 0.25 0.025 0.8231890 1.0000000 8
11 0.275 0.025 0.8044747 0.9806156 7
12 0.3 0.025 0.8044747 1.0000000 7
13 0.325 0.025 0.5193348 0.6852822 6
14 0.35 0.025 0.5193348 1.0000000 6
15 0.375 0.025 0.5192552 0.9998965 5
16 0.4 0.050 0.5192552 1.0000000 5
17 0.425 0.050 0.5192552 1.0000000 5
18 0.45 0.025 0.3397770 0.7370673 3
19 0.475 0.275 0.3397770 1.0000000 3
20 0.5 0.275 0.3397770 1.0000000 3
21 0.525 0.275 0.3397770 1.0000000 3
22 0.55 0.275 0.3397770 1.0000000 3
23 0.575 0.275 0.3397770 1.0000000 3
24 0.6 0.275 0.3397770 1.0000000 3
25 0.625 0.275 0.3397770 1.0000000 3
26 0.65 0.275 0.3397770 1.0000000 3
27 0.675 0.275 0.3397770 1.0000000 3
28 0.7 0.275 0.3397770 1.0000000 3
29 0.725 0.275 0.3397770 1.0000000 3
30 0.75 0.025 0.3278743 0.9786017 2
31 0.775 0.250 0.3278743 1.0000000 2
32 0.8 0.250 0.3278743 1.0000000 2
33 0.825 0.250 0.3278743 1.0000000 2
34 0.85 0.250 0.3278743 1.0000000 2
35 0.875 0.250 0.3278743 1.0000000 2
36 0.9 0.250 0.3278743 1.0000000 2
37 0.925 0.250 0.3278743 1.0000000 2
38 0.95 0.250 0.3278743 1.0000000 2
39 0.975 0.250 0.3278743 1.0000000 2
40 1 0.250 0.3278743 1.0000000 2
scGPS results
> nor2_scGPS_cluster$optimalClust$KeyStats
Height Stability RandIndex ConsecutiveRI Cluster_count
1 0.025 0.250 0.9973368 1.0000000 10
2 0.05 0.250 0.9973368 1.0000000 10
3 0.075 0.250 0.9973368 1.0000000 10
4 0.1 0.250 0.9973368 1.0000000 10
5 0.125 0.250 0.9973368 1.0000000 10
6 0.15 0.250 0.9973368 1.0000000 10
7 0.175 0.250 0.9973368 1.0000000 10
8 0.2 0.250 0.9973368 1.0000000 10
9 0.225 0.250 0.9973368 1.0000000 10
10 0.25 0.250 0.9973368 1.0000000 10
11 0.275 0.025 0.9592107 0.9617593 8
12 0.3 0.025 0.9592107 1.0000000 8
13 0.325 0.025 0.6887011 0.7256560 7
14 0.35 0.025 0.6887011 1.0000000 7
15 0.375 0.025 0.6566005 0.9640588 6
16 0.4 0.075 0.6566005 1.0000000 6
17 0.425 0.075 0.6566005 1.0000000 6
18 0.45 0.075 0.6566005 1.0000000 6
19 0.475 0.025 0.6373315 0.9779964 5
20 0.5 0.025 0.6373315 1.0000000 5
21 0.525 0.025 0.4967156 0.8281841 4
22 0.55 0.150 0.4967156 1.0000000 4
23 0.575 0.150 0.4967156 1.0000000 4
24 0.6 0.150 0.4967156 1.0000000 4
25 0.625 0.150 0.4967156 1.0000000 4
26 0.65 0.150 0.4967156 1.0000000 4
27 0.675 0.150 0.4967156 1.0000000 4
28 0.7 0.025 0.4961135 0.9941510 3
29 0.725 0.100 0.4961135 1.0000000 3
30 0.75 0.100 0.4961135 1.0000000 3
31 0.775 0.100 0.4961135 1.0000000 3
32 0.8 0.100 0.4961135 1.0000000 3
33 0.825 0.025 0.3896436 0.8462418 2
34 0.85 0.175 0.3896436 1.0000000 2
35 0.875 0.175 0.3896436 1.0000000 2
36 0.9 0.175 0.3896436 1.0000000 2
37 0.925 0.175 0.3896436 1.0000000 2
38 0.95 0.175 0.3896436 1.0000000 2
39 0.975 0.175 0.3896436 1.0000000 2
40 1 0.175 0.3896436 1.0000000 2
We have developed a new version of clustering that is implemented in scGPS, with the new name: SCORE (Stable Clustering at Optimal Resolution). The ascend clustering has been updated as well, so you may want to install the latest version and check the results. Cheers
I did use the latest version of ascend and scGPS to conduct my analysis.
In ascend, I used runCORE
function. In scGPS, I used CORE_scGPS
function.
other attached packages:
[1] scGPS_0.9.9 dynamicTreeCut_1.63-1
[3] ascend_0.9.6 SingleCellExperiment_1.4.0
I did clustering analysis by ascend package and have EMSet object for my data. How can I use it for analysis by scGPS? Thanks.