Table 1. Clusters categorised based on over- or under-representation of the various serum and growth conditions.

A
Factor	Enrichment	Actual	Expected	P-value
Cluster 1, n = 16
HS	Positive	16	4	<0.001
FSC	Negative	0	4	<0.001
DMEM	Negative	0	4	<0.001
CS-FCS	Negative	0	4	<0.001
Cluster 2 n = 37
FCS	Positive	16	9.25	<0.001
HS	Negative	3	9.25	<0.001
Cluster 3 n = 18
No serum enrichment
Cluster 4 n = 13
DMEM	Positive	10	3.25	<0.001
HS	Negative	0	3.25	<0.01
B
Cluster 1, n = 16
No enrichment for treatment
Cluster 2 n = 37
Control	Positive	20	12.3	<0.001
Vitamin D	Negative	5	12.3	<0.001
Cluster 3 n = 18
Vitamin D	Positive	18	6	<0.001
Insulin	Negative	0	6	<0.001
Control	Negative	0	6	<0.001
Cluster 4 n = 13
Insulin	Positive	10	4.3	<0.001
Vitamin D	Negative	0	4.3	<0.001

For each cluster the number of observations belonging to each category was calculated (‘Actual’) and was compared with the expected number of observations in each category if the observations would have been evenly distributed (‘Expected’) (i.e hypergeometrical distribution). P-values were calculated with fishers exact test (p-value). A) cover enrichment of serum-conditions in the different clusters B) concern enrichment of treatment conditions in the different clusters.