Regarding the issue of n.cells.total.g1 and n.cells.total.g2 in 10x single-cell sequencing.

[QingCheZhang]

I tried the example code in https://wenweixiong.github.io/MARVEL_Droplet.html, and i have some questions about this code

marvel <- CompareValues.Genes.10x(MarvelObject=marvel,
                                  show.progress=FALSE
                                  )

head(marvel$DE$SJ$Table)

##           coord.intron gene_short_name n.cells.total.g1 n.cells.expr.sj.g1
## 1 chr1:1373903:1373999        AURKAIP1             5937               5858
## 2 chr1:1402257:1405808          MRPL20             5937               2955
## 3 chr1:1405887:1406908          MRPL20             5937               1069
## 4 chr1:2189782:2193638          FAAP20             5937               4686
## 5 chr1:6197757:6199561           RPL22             5937               4839
## 6 chr1:6820251:6825091          CAMTA1             5937                479
##   pct.cells.expr.sj.g1 n.cells.expr.gene.g1 pct.cells.expr.gene.g1
## 1                98.67                 5897                  99.33
## 2                49.77                 5845                  98.45
## 3                18.01                 5845                  98.45
## 4                78.93                 5479                  92.29
## 5                81.51                 5937                 100.00
## 6                 8.07                 5753                  96.90
##   sj.count.total.g1 gene.count.total.g1 psi.g1 n.cells.total.g2
## 1             40782               49806  81.88             5937
## 2              4425               36002  12.29             5937
## 3              1199               36002   3.33             5937
## 4             10815               19491  55.49             5937
## 5             11697              193189   6.05             5937
## 6               498               27599   1.80             5937
##   n.cells.expr.sj.g2 pct.cells.expr.sj.g2 n.cells.expr.gene.g2
## 1               5798                97.66                 5858
## 2               3516                59.22                 5859
## 3               1249                21.04                 5859
## 4               5270                88.77                 5715
## 5               5499                92.62                 5937
## 6                675                11.37                 5717
##   pct.cells.expr.gene.g2 sj.count.total.g2 gene.count.total.g2 psi.g2
## 1                  98.67             32201               38876  82.83
## 2                  98.69              5754               35035  16.42
## 3                  98.69              1445               35035   4.12
## 4                  96.26             15551               25371  61.29
## 5                 100.00             19280              189020  10.20
## 6                  96.29               721               27367   2.63
##       log2fc delta pval mean.expr.gene.norm.g1.g2 n.cells.total.norm.g1
## 1 0.01644263  0.95    0                  2.138988                  5937
## 2 0.39040352  4.13    0                  1.901503                  5937
## 3 0.24177679  0.79    0                  1.901503                  5937
## 4 0.14100507  5.80    0                  1.410974                  5937
## 5 0.66780357  4.15    0                  4.074651                  5937
## 6 0.37454272  0.83    0                  1.631310                  5937
##   n.cells.expr.gene.norm.g1 pct.cells.expr.gene.norm.g1 mean.expr.gene.norm.g1
## 1                      5897                       99.33               2.501344
## 2                      5845                       98.45               2.108606
## 3                      5845                       98.45               2.108606
## 4                      5479                       92.29               1.436871
## 5                      5937                      100.00               4.350727
## 6                      5753                       96.90               1.821909
##   n.cells.total.norm.g2 n.cells.expr.gene.norm.g2 pct.cells.expr.gene.norm.g2
## 1                  5937                      5858                       98.67
## 2                  5937                      5859                       98.69
## 3                  5937                      5859                       98.69
## 4                  5937                      5715                       96.26
## 5                  5937                      5937                      100.00
## 6                  5937                      5717                       96.29
##   mean.expr.gene.norm.g2 log2fc.gene.norm pval.gene.norm pval.adj.gene.norm
## 1               1.776632      -0.72471240   0.000000e+00       0.000000e+00
## 2               1.694400      -0.41420606   0.000000e+00       0.000000e+00
## 3               1.694400      -0.41420606   0.000000e+00       0.000000e+00
## 4               1.385077      -0.05179485   8.407605e-13       9.278512e-13
## 5               3.798576      -0.55215116   0.000000e+00       0.000000e+00
## 6               1.440711      -0.38119821  8.205003e-297      1.623418e-296

I wonder why n.cells.total.g1 and n.cells.total.g2 are both 5937 when Group 1 has 11244 cells.

thanks

[szhaoaf]

The reason is that that you set "downsample=TRUE". Then it will downsample to select the minimal number of group between these two groups.

See the wrapped codes:

if (downsample == TRUE) {
set.seed(seed)
n.cells.downsample <- min(length(cell.group.g1), length(cell.group.g2))
cell.group.g1 <- sample(cell.group.g1, size = n.cells.downsample,
replace = FALSE)
cell.group.g2 <- sample(cell.group.g2, size = n.cells.downsample,
replace = FALSE)
}

[wenweixiong]

You're absolute right! We downsample and match the cell numbers in both groups to unsure that any differences in splice junction usage/expression are not due to differences in the sample size of the two groups.