hcluster_matrix.sh error about get_homologues

[eead-csic-compbio]

Hi Aditya, can you please provide your ANI matrix, with annonymized labels if you prefer?
Bruno

[aupadrasta]

I have updated the software. now its giving an error
Error in library(dendextend) : there is no package called 'dendextend'
Calls: suppressPackageStartupMessages -> withCallingHandlers -> library
Execution halted

ERROR: File hclust_gower-ward.D2_FperidonATCC33693PP4_f0_0taxa_algOMCL_e0_Avg_identity_tree.pdf was NOT generated!
File gower_dist_matrix.tab was generated
ERROR: File hclust_gower-ward.D2_FperidonATCC33693PP4_f0_0taxa_algOMCL_e0_Avg_identity_heatmap.pdf was NOT generated!
ERROR: File hclust_gower-ward.D2_FperidonATCC33693PP4_f0_0taxa_algOMCL_e0_Avg_identity_tree.ph was NOT generated!
File was generated
File was generated

[eead-csic-compbio]

Can you please share the command line you used? After updating I guess now you're using hcluster_pangenome_matrix, right? This script was just updated so you might have found the first bug of this version. Pablo and I will look into it in the next 1-2 days, ok? Thanks for the report, Bruno

[aupadrasta]

Yes
I have used the
hcluster_pangenome_matrix.sh -i
Which i have sent to you

[eead-csic-compbio]

Hi,
I just had a quick check and it seems that your ANI matrix contains a lot of NA cells, which are most likely causing this error. Is this because this ANI was obtained from a set of genomes with core size = 0?

[aupadrasta]

Yes, the core size = 0. I have just rechecked it to confirm.
There should be at least >20% identity among all the sequences which were subjected to analysis. I have no clue why its giving as NA??

[eead-csic-compbio]

Hi again,
well, those are pretty low identities, well into the twilight zone.
Please note that coverage is probably the most important parameter and you may have to relaxed to recover those BLAST hits. Finally, even if BLAST gets them it does not mean that they will be called orthologues.
Take home message: you need a core genome set of genes to compute ANI

[aupadrasta]

Dear Bruno,
Would you suggest any other way to do the analysis? or any changes to be made to run the scripts?
Thanks
Aditya

[eead-csic-compbio]

You have several options:

1. remove poor assemblies which might be missing core genes
2. remove species which are too distant to your genomes of interest
3. use get_homologues.pl -C 50 or even lower as remote homologues seem to display lower align coverage than close sequences

[vinuesa]

Hi Aditya, there are two problems:

1. The absolutely critical one is that all your rows contains NAs. R cannot proceed with such a matrix. As suggested by Bruno, try to relax the coverage parameter (leave it at -C 0). There is nothing to compute if the core size = 0. Another option is to compute the Avg_identity (ANI matrix) using the pan-genome instead.

2. You will also need to install the dendextend package, a new requirement of version 1.0.* of plot_matrix_heatmap.sh

Please let us know if we can further assist you with these matters.

[vinuesa]

Just added the complete.cases(tab)check, at the head of the script to stop its execution if( dim(tab)[1] < 3 ), as in your case.