Inference of the pan‐ncRNAome - labbces/sugarcane_RNAome GitHub Wiki

Inference of the pan‐ncRNAome

We have developed a pipeline to infer the pan-ncRNAome. The pipeline follows the steps outlined below:

Convert the MMseqs2 similarity list into a matrix similar to OrthoFinder

This step transforms the format of the cluster file generated by MMseqs2 into a format identical to OrthoFinder's clustering. Each cluster's name is listed on the rows, and the genotypes are represented in columns, with each cell containing the sequence ID of the genotype within each cluster. Like this example.

This step is performed with this script.

Note: Before running this script, it is necessary to carry out an intermediate step to keep only the genotype names in the second column of the clusters file generated by MMseqs2, using the following code:

awk -F '\t' 'BEGIN {OFS = FS} {split($2, arr, "_"); $2 = arr[1]; print}' DB_clust.tsv >> DB_clust_genotypeName.tsv 

Before (MMseqs2 clustering output format):

Co06022_k31_TRINITY_DN17309_c0_g2_i1 Co06022_k31_TRINITY_DN17309_c0_g2_i1
Co06022_k25_TRINITY_DN6264_c0_g2_i8 Co06022_k25_TRINITY_DN6264_c0_g2_i8
Co06022_k25_TRINITY_DN6264_c0_g2_i8 Co06022_k25_TRINITY_DN6264_c0_g2_i10
Co06022_k25_TRINITY_DN6264_c0_g2_i8 Co06022_k25_TRINITY_DN6264_c0_g2_i4
Co06022_k25_TRINITY_DN6264_c0_g2_i8 Co06022_k31_TRINITY_DN12248_c0_g2_i5

intermediate step (MMseqs2 clustering output format with only genotype names):

Co06022_k31_TRINITY_DN17309_c0_g2_i1 Co06022
Co06022_k25_TRINITY_DN6264_c0_g2_i8 Co06022
Co06022_k25_TRINITY_DN6264_c0_g2_i8 Co06022
Co06022_k25_TRINITY_DN6264_c0_g2_i8 Co06022
Co06022_k25_TRINITY_DN6264_c0_g2_i8 Co06022

After (OrthoFinder like matrix):

Orthogroup B1 B2 CP74-2005 Co06022 group_name
OG1 0 0 0 1 OG1
OG2 0 0 0 4 OG2

Manually remove the last column (group_name) from the file DB_clust_groups.tsv

This step is necessary to avoid including duplicate information regarding the cluster names. See this example.

cut -f1-49 DB_clust_groups.tsv >> DB_clust_groups_withoutLastColumn.tsv

Calculate pan-ncRNAome classes (pan-, accessory-, soft-core-, hard-core and exclusive-ncRNAome's)

We developed a python script that categorizes each cluster into the following classes:

  • Pan (sum of total classes)
  • Hard-core (clusters with 100% of genotypes present)
  • Soft-core (clusters with 80% of genotypes present)
  • Exclusive (clusters with only one genotype present)
  • Accessory (clusters with more than one genotype up to 80% of genotypes present)

This bash script was used to run the script to calculate the pan-ncRNAome classes from the above reformatted similarity matrix, generating this file.

Output format example:

Genotypes Groups Class Sequences
1 96872 pan 120200
1 0 accessory 0
1 0 soft-core 0
1 96872 hard-core 120200
1 3310316 exclusive 0

Run the script to plot the pan-ncRNAome

This script generates strip plots of the group classes, with the number of identified groups represented on the Y-axis and the number of genotypes on the X-axis.

At the end of the pipeline, we obtain a representation of the group classes as follows:

pan_ncRNAome.png

Core groups and transcripts

Groups

ncRNA_core_groups.PNG

Transcripts

ncRNA_core_transcripts.PNG

Length distribution and GC content (300bp+).

B1_transcriptome.cpc_ncrnas_stats_histograms.png

Pan-transcriptome taxonomy

48_transcriptomes.report.krona.png