Add notation of the output

R/Gibbs_sampler.R

@@ -346,6 +346,8 @@ SpaTopic_inference<-function(tissue, ntopics, sigma = 50, region_radius = 400, k
   }
 
   ###  release memory for large items
+  celltype<-levels(itr_df$type)
+  cellname<-rownames(itr_df)
   rm(list= c("itr_df"))
   gc()
 
@@ -371,7 +373,8 @@ SpaTopic_inference<-function(tissue, ntopics, sigma = 50, region_radius = 400, k
   if(ninit > 1){ ## need update only when several initialization
 
       ## number of regions
-    M<-max(D)+1  ## M might be changed since D changed
+    M<-max(D)+1  ## M might be changed since D changed 
+    # A potential bug (what if the max(D) has been changed after gibbs sampling)
 
     Ndk <- table_2d_fast(D, Z, M, K) ## number of cells per topic per region (image specific)
     Nwk <- table_2d_fast(C, Z, V, K) ## number of cells per topic per celltype
@@ -408,6 +411,16 @@ SpaTopic_inference<-function(tissue, ntopics, sigma = 50, region_radius = 400, k
   message("Output model perplexity:")
   message(gibbs.res$Perplexity)
 
+  ## Beta: Topic Content
+  gibbs.res$Beta<-as.data.frame(gibbs.res$Beta)
+  colnames(gibbs.res$Beta)<-paste0("topic",1:K)
+  rownames(gibbs.res$Beta)<-celltype
+
+  ## Z.trace: posterior prob for every single cell
+  gibbs.res$Z.trace<-as.data.frame(gibbs.res$Z.trace/niter)
+  colnames(gibbs.res$Z.trace)<-paste0("topic",1:K)
+  rownames(gibbs.res$Z.trace)<-cellname
+
   return(gibbs.res)
 }
 

README.md

@@ -69,15 +69,29 @@ It is stored in a Seurat v5 object.
 
 ``` r
 str(gibbs.res)
-#List of 8
-# $ Perplexity   : num 11.3
-# $ Deviance     : num 485960
-# $ loglikelihood: num -242980
-# $ Beta         : num [1:38, 1:7] 0.03587 0.02539 0.00755 0.01858 0.02585 ...
-# $ Theta        : num [1:971, 1:7] 0.855601 0.000232 0.999269 0.99889 0.998725 ...
-# $ Ndk          : int [1:971, 1:7] 107 0 82 54 47 72 100 0 0 0 ...
-# $ Nwk          : int [1:38, 1:7] 390 276 82 202 281 505 697 522 29 58 ...
-# $ Z.trace      : int [1:100149, 1:7] 13 173 27 164 157 4 20 21 15 19 ...
+# List of 8
+#  $ Perplexity   : num 11.3
+#  $ Deviance     : num 485960
+#  $ loglikelihood: num -242980
+#  $ Beta         :'data.frame':	38 obs. of  7 variables:
+#   ..$ topic1: num [1:38] 0.03587 0.02539 0.00755 0.01858 0.02585 ...
+#   ..$ topic2: num [1:38] 6.51e-03 3.55e-02 2.62e-06 5.80e-04 7.75e-01 ...
+#   ..$ topic3: num [1:38] 4.54e-06 4.54e-06 9.13e-04 3.45e-01 1.73e-03 ...
+#   ..$ topic4: num [1:38] 0.02664 0.01743 0.00186 0.0152 0.08919 ...
+#   ..$ topic5: num [1:38] 2.99e-06 2.99e-06 5.32e-03 1.91e-02 4.90e-03 ...
+#   ..$ topic6: num [1:38] 6.35e-06 6.35e-06 2.04e-02 3.43e-03 6.35e-06 ...
+#   ..$ topic7: num [1:38] 0.00534 0.00699 0.00604 0.01843 0.00655 ...
+#  $ Theta        : num [1:971, 1:7] 0.855601 0.000232 0.999269 0.99889 0.998725 ...
+#  $ Ndk          : int [1:971, 1:7] 107 0 82 54 47 72 100 0 0 0 ...
+#  $ Nwk          : int [1:38, 1:7] 390 276 82 202 281 505 697 522 29 58 ...
+#  $ Z.trace      :'data.frame':	100149 obs. of  7 variables:
+#   ..$ topic1: num [1:100149] 0.065 0.865 0.135 0.82 0.785 0.02 0.1 0.105 0.075 0.095 ...
+#   ..$ topic2: num [1:100149] 0 0 0 0 0 0 0 0 0 0 ...
+#   ..$ topic3: num [1:100149] 0.275 0.005 0.21 0.005 0.005 0.77 0.02 0.015 0.085 0.075 ...
+#   ..$ topic4: num [1:100149] 0.415 0 0 0.01 0.005 0.1 0.665 0.62 0.015 0.025 ...
+#   ..$ topic5: num [1:100149] 0.005 0.01 0 0 0 0 0.005 0.005 0.005 0 ...
+#   ..$ topic6: num [1:100149] 0 0 0.655 0.165 0.205 0.005 0 0 0 0 ...
+#   ..$ topic7: num [1:100149] 0.24 0.12 0 0 0 0.105 0.21 0.255 0.82 0.805 ...
 ```
 
 ## Citation