Table 2 Overview of pairwise differential analysis of spatial transcriptomics (ST), bulk transcriptomics (BT) and bulk methylation (BM) data

From: Spatial transcriptomics reveals strong association between SFRP4 and extracellular matrix remodeling in prostate cancer

Statistics

Data

Sample type

Variables

Group 1

Group 2

Quasi-likelihood negative binomial generalized log-linear model

ST

Spot

2436

Normal stroma (n = 3186)

Cancer stroma (n = 4802)

Quasi-likelihood negative binomial generalized log-linear model

ST

Spot

2436

LG Cancer stroma (n = 2620)

HG Cancer stroma (n = 2182)

Quasi-likelihood negative binomial generalized log-linear model

ST

Spot

2436

Relapse-free cancer stroma (n = 1469)

Relapse cancer stroma (n = 3333)

Quasi-likelihood negative binomial generalized log-linear model

ST

Spot

2436

Relapse-free LG cancer stroma (n = 1219)

Relapse LG cancer stroma (n = 1401)

Quasi-likelihood negative binomial generalized log-linear model

ST

Spot

2436

Relapse-free HG cancer stroma (n = 250)

Relapse HG cancer stroma (n = 1932)

Linear mixed models

BT

Bulk tissue

1

Non-cancer (n = 61)

Cancer (n = 115)

Linear mixed models

BT

Bulk tissue

1

Relapse-free (n = 49)

Relapse (n = 127)

Linear mixed models

BT

Bulk tissue

1

LG cancer (n = 61)

HG cancer (n = 54)

Linear mixed models

BT

Bulk tissue

1

Relapse-free cancer (n = 29)

Relapse cancer (n = 86)

Linear mixed models

BM

Bulk tissue

20

Non-cancer (n = 29)

Cancer (n = 35)

Linear mixed models

BM

Bulk tissue

20

Relapse-free (n = 20)

Relapse (n = 44)

Linear mixed models

BM

Bulk tissue

20

LG cancer (n = 17)

HG cancer (n = 10)

Linear mixed models

BM

Bulk tissue

20

Relapse-free cancer (n = 14)

Relapse cancer (n = 32)

  1. For each statistical test the input sample was either a ST spot (spatial data) or whole tissue sample (bulk data). Samples within the same group being tested against another group were considered biological replicates.
  2. LG low-grade, HG, high-grade.