search

satijalab / seurat

R toolkit for single cell genomics
http://www.satijalab.org/seurat
Other
2.3k stars 917 forks source link

Differential expressed genes from merged seurat objects #2351

Closed tulikakakati closed 4 years ago

tulikakakati commented 4 years ago

Hi, Please correct me, if I am wrong. I have two different samples of scRNAs (Normal and Disease). First I created two seurat objects (n and d) and then merged them using merge(n,d). I used the merged_object further for differential expression analysis after clustering. I used FindMarkers(merged_object, ident.1 = id, logfc.threshold = 0.25, test.use = "roc", only.pos = TRUE) function to find the marker genes for each cluster represents the differentially expressed genes.

My queries to you is as follows:

The cell clusters composed of cells from both normal and disease samples. So do the markers we found from these clusters signify the differentially expressed genes between normal and disease cells or between different clusters of merged objects. Thank you

timoast commented 4 years ago

It depends what groups you have contrasted in FindMarkers, from this it's not clear what the variable id is. Does it group cells by cluster and normal/disease, or only cluster, or only normal/disease?

tulikakakati commented 4 years ago

Here, id is each cell cluster.

It grouped cells by cluster with normal/ disease. For example, cluster 1 will have cells from both normal as well as disease samples. ah-ACACGCGCACGGCGTT ah-ACTATGGCATAACTCG ah-AGATAGAGTTTCCAAG ah-CCTGTTGCAATAGTCC ah-GCATGATCAGTCAACT ah-GGTAATCTCATCACAG ah-GTGGCGTTCTCGCCTA ah-GTTCCGTTCAAAGGAT ah-TCATTGTCACTCCCTA ah-TCCTTTCCACGGTCTG ah-TTCTTGAAGATTGCGG hc-AACCCAAAGGCGTTGA hc-AATTCCTGTGACTAAA hc-AGATGAAGTTGGCCGT hc-ATAGGCTAGGCTCTCG hc-ATCGATGAGTCCGTCG hc-ATGACCAAGCTCAGAG hc-ATGACCATCACCTGGG hc-ATTCCTAGTGGCGTAA hc-CAGATTGTCGGTTAGT hc-CCACACTAGATCCGAG hc-CCTAACCTCCACTGAA hc-CCTACGTGTAGCTTGT hc-CCTCAACTCCTACCGT hc-GGCTTTCAGGATCATA hc-GTACAACTCTCCTGTG

And the marker gene I found from this cell cluster is listed below:

myAUC avg_diff power pct.1 pct.2  
LTB 0.871 1.117306 0.742 0.995 0.666
IL7R 0.84 0.993148 0.68 0.944 0.489
RPS29 0.821 0.594928 0.642 1 0.997
TPT1 0.813 0.509112 0.626 1 0.999
IL32 0.791 0.753823 0.582 0.955 0.612
RPL41 0.781 0.389246 0.562 1 0.999
RPS27 0.779 0.483321 0.558 1 0.999
RPL36 0.772 0.423398 0.544 1 0.994
RPL39 0.771 0.416769 0.542 1 0.998
RPL23A 0.769 0.424361 0.538 1 0.998
RPS18 0.766 0.453968 0.532 1 0.999
RPL37 0.765 0.387516 0.53 1 0.999
RPS21 0.755 0.410167 0.51 0.998 0.969
RPL18A 0.754 0.387009 0.508 1 0.999
EEF1A1 0.751 0.376171 0.502 1 0.999
LDHB 0.75 0.487094 0.5 0.944 0.671
RPL13 0.75 0.398402 0.5 1 0.999
RPL14 0.746 0.366663 0.492 1 0.998
CD3D 0.741 0.470769 0.482 0.945 0.556
RPS28 0.74 0.340005 0.48 1 0.999
RPL7A 0.74 0.320323 0.48 1 0.999
RPS4Y1 0.739 0.541711 0.478 0.893 0.611
AQP3 0.739 0.531384 0.478 0.598 0.134
RPSA 0.739 0.411218 0.478 1 0.99
RPL10A 0.739 0.384066 0.478 1 0.997
RPLP1 0.739 0.345047 0.478 1 0.999
RPS2 0.738 0.356415 0.476 1 0.999
RPS12 0.735 0.385934 0.47 1 0.999
timoast commented 4 years ago

If the groups you're contrasting contain cells from both conditions then the answer is no. If you want to contrast disease/normal you will need to add that as a grouping variable and specify the grouping variable in FindMarkers. See this vignette for more information: https://satijalab.org/seurat/v3.1/immune_alignment.html

tulikakakati commented 4 years ago

Thank you