3. Understanding the CluBCpG output data

3.1. Coverage output

The output from clubcpg-coverage looks like this:

clubcpg-coverage output

bin_id

n_reads

n_cpgs

chr19_61300

4

1

chr19_61400

4

2

chr19_89800

25

4

Each row represents 1 bin.

Note

A header row is shown on this table for clarity, however the real csv file generated will not have a header row.

  • bin_id

    Represents the unique bin in the genome. The underscore character (__) separates the chromosome and the genomic coordinate. The genomic coordiate represents the end-point of a bin.

    • ex: If --bin_size was set to 100, chr19_61300 would represent chr19:61200-61300

  • n_reads

    Number of reads which fully cover all CpGs within the bin

  • n_cpgs

    Number of CpGs within the bin

3.2. Cluster output

The output from clubcpg-cluster looks like this:

clubcpg-cluster output

bin_id

input_label

methylation

class_label

read_number

cpg_number

cpg_pattern

class_split

chr2_10700

AB

0.8333

0

6

6

1;1;1;1;1;0

A=5,B=1

chr2_10700

A

1

1

12

6

1;1;1;1;1;1

A=12

Each row represents 1 cluster.

  • bin_id

    Represents the unique bin in the genome. The underscore character (__) separates the chromosome and the genomic coordinate. The genomic coordiate represents the end-point of a bin.

  • input_label

    Represents the input file this cluster was found in

    • Single-file mode
      • This will show the BAM file name specified with the -a flag.

    • Two-file mode:
      • A = Input BAM specified in the -a flag

      • B = Input BAM specified in the -b

      • AB = found in both input BAM files

  • methylation

    The methylation level of the CpG pattern found in this cluster

  • class_label

    A unique identifier for each cluster found within a given bin (only unique within a bin)

  • read_number

    The number of reads found within this cluster

  • cpg_number

    The number of CpGs in this cluster

  • cpg_pattern

    The methylation pattern of the cluster

    • 1 = methylated

    • 0 = unmethylated

  • class_split

    The breakdown of how many reads in a given cluster came from each input BAM.

    • Useful for AB clusters.

    • If cluster is A or B, this number will match the read_number column.

Warning

If clustering was performed with --remove_noise False (this is True by deafult) you may find clusters with a class_label of -1.

These represents noise reads, which are reads where a given CpG patterns was only observed ONCE.

The read_number column may be of interest to you, but the other columns do NOT represent true values. They will only represent one of the noise patterns found. These values may be set to null in future versions of CluBCpG.