Alakazam issue

[evan.kransdorf]

Hello All,

I am finding an issue in Alakazam when doing lineage analysis.

I am getting the following error:

"Error in buildPhylipLineage(clone, phylip_exec, rm_temp = TRUE)"

"The germline and input sequences are not the same length for clone x"

I worked through this and looks like the error arises because within the clone I have created the sequence length and Germline sequence length are not the same. It looks like they are one nt off.

Any thoughts about why this is occurring? I assume it's an issue from Change-O?

Many thanks!

Evan

[Jason Vander Heiden]

Greetings Evan,

It could be a bug in changeo. When you ran CreateGermlines did you use the `--cloned` argument?  This will assign the exact same germline to every sequence in the entire clone. CreateGermlines should fail sequences where the length of the germline and observed sequence don't match, but I can imagine cases were the clonal consensus germline differs in length from one of the sequences with the clone.

That said, if the missing nucleotides are at the end of the sequences, then  you should be able to fix it by adding the argument  `pad_end=TRUE` to the `makeChangeoClone` call for the clone. This will add an appropriate number of N characters to the end of each sequence until they are all the same length (this happens a lot with 454 data).

--
You received this message because you are subscribed to the Google Groups "immcantation-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to immcantation-us...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/immcantation-users/6036a6aa-aea1-4ada-b195-b15e542a3e42o%40googlegroups.com.

[evan.kransdorf]

Hello Jason,

Thanks for your reply. Indeed we used the --cloned argument. I also tried the pad_end=TRUE but it did not help. It may be a bug. 

Also, while I have you, any thoughts about the maximum clone size? My computer is frequently running out of memory for large clones (20k sequences).

Thanks again! Evan

On Thursday, July 16, 2020 at 9:28:44 AM UTC-7, Jason Vander Heiden wrote:

Greetings Evan,

It could be a bug in changeo. When you ran CreateGermlines did you use the `--cloned` argument?  This will assign the exact same germline to every sequence in the entire clone. CreateGermlines should fail sequences where the length of the germline and observed sequence don't match, but I can imagine cases were the clonal consensus germline differs in length from one of the sequences with the clone.

That said, if the missing nucleotides are at the end of the sequences, then  you should be able to fix it by adding the argument  `pad_end=TRUE` to the `makeChangeoClone` call for the clone. This will add an appropriate number of N characters to the end of each sequence until they are all the same length (this happens a lot with 454 data).

On Jul 15, 2020, at 6:10 AM, evan.kransdorf <evan.k...@gmail.com> wrote:

Hello All,

I am finding an issue in Alakazam when doing lineage analysis.

I am getting the following error:

"Error in buildPhylipLineage(clone, phylip_exec, rm_temp = TRUE)"

"The germline and input sequences are not the same length for clone x"

I worked through this and looks like the error arises because within the clone I have created the sequence length and Germline sequence length are not the same. It looks like they are one nt off.

Any thoughts about why this is occurring? I assume it's an issue from Change-O?

Many thanks!

Evan

--
You received this message because you are subscribed to the Google Groups "immcantation-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to immcantation-users+unsub...@googlegroups.com.

[Jason Vander Heiden]

Hi Evan,

Would you be able to send us (immcan...@googlegroups.com) a tsv of the offending clone? We can take a look at it.

20K sequence in a clone is huge. Is that unique sequences? I don't know if there's way to improve phylip's memory utilization (I doubt it), but you could randomly subsample the giant clone to something like 1K sequences.

To unsubscribe from this group and stop receiving emails from it, send an email to immcantation-us...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/immcantation-users/41ad8bf5-6de5-4bb9-b362-5b7a76b90031o%40googlegroups.com.