count.tsv incomplete

[pizzabytes]

Hi,

The count.tsv isn't showing all the entries from the .bed file with the 3' UTR coordinates (for example, some chromosomes are missing entirely) - where might this issue be stemming from?

Thanks!

[t-neumann]

Hm do you have some duplicate entries in there? Can you check?
Also, can you share a few lines of the bed file so I could have a look what potentially is going on?

[pizzabytes]

Hi @t-neumann,

I ended up switching to use ngm to map since I had paired end reads, and the resulting count.tsv reflects the .bed file. I then proceeded with slamdunk filter, slamdunk snp, and slamdunk count with no issues, followed by alleyoop on Galaxy. However, I am not seeing a T>C excess on the plus strand nor an A>B excdess on the minus strand.

I followed the steps from your response here.

Essentially, the script was like this:

ngm
-1 /path/to/sample1_R1.fastq.gz
-2 /path/to/sample1_R2.fastq.gz
-r /path/to/reference.fasta
-o /path/to/sample1_ngm.bam
-b
-t 8
--slam-seq 2
--rg-id 1
--rg-sm sample1:pulse:0

Is the the correct way to map paired end data to use in the slamdunk pipeline? I am trying to figure out if the lack of a T>C excess is due to an error I am making in the pipeline or if the labelling itself failed.

Thank you!

[t-neumann]

Yes that looks correct to me. I would also try to map only the first and second read separately and see if you see an increase in T>C reads in one or the other condition to see if paired-end T>C reads are treated correctly.

[pizzabytes]

Hi @t-neumann ,

I have a new set of samples which showed T>C conversions in the R2 reads after running Slamdunk and Alleyoop. The R1 reads show no increase in T>C conversions but an increase in A>G conversions when looking at the plus strand. The inverse occurs in the minus strand. Since my library prep was stranded, I realised that R1 should be reverse complemented first, which was also suggested in #149 . Then, I should be able to continue with the code above with no issue.

I was wondering how these stranded, paired-end samples can be processed with Slamdunk and Alleyoop on Galaxy? Since there is no paired-end option, would reverse complementing R1 reads followed by interleaving with R2 reads be analogous?

Thank you!

[t-neumann]

I would actually use NGM for paired end read mapping as you did, but only then use the R1 reads - otherwise you might overcount (two reads from a given fragment) certain transcripts and also get some artifacts (if one read mate maps to gene A and the other e.g. to gene B).

Alternatively, I guess the better solution for paired-end reads is to follow suggestions in this thread:

#165