Multiple fasta sequences, but of different lengths

[SheepwormJM]

Hi,

I think I may have an idea of the right answer to my question, but hoping to check I'm not going to do something stupid. I'm wondering if I need to create an multisequence alignment file rather than using a multifasta. My ultimate aim is to produce a haplotype network.

I have a locus which I've amplified, but which has lots of small indels within an intronic part of the sequence.

I have used the following code to load a multifasta file into R, but cannot convert it to a matrix because the sequences are uneven lengths. I can't see any way to get a matrix to load with 'NA' for gaps, and I'm not sure if pegas would subsequently re-align the sequences, or assume they were aligned, if the matrix stuck a lot of NAs on the end of each sequence, rather than internally.

library("apex")
library("adegenet")
library("pegas")
library("mmod")
library("poppr")

# To get a SINGLE fasta file in: 
myseq<-read.FASTA("ASV_multifasta.fa")
myseq # Provides the summary information of the file

2172 DNA sequences in binary format stored in a list.

Mean sequence length: 329.453 
   Shortest sequence: 294 
    Longest sequence: 350 

Labels:
ASV3 BO_04_M
ASV3 BO_04_M
ASV3 BO_04_M
ASV3 BO_04_M
ASV3 BO_04_M
ASV3 BO_04_M
...

Base composition:
    a     c     g     t 
0.282 0.204 0.190 0.324 
(Total: 715.57 kb)

# We need to make it as a matrix:
myseqmatrix<-as.matrix(myseq)

Then I get the error telling me it won't work because the sequences are different lengths.

Error in as.matrix.DNAbin(myseq) : 
  DNA sequences in list not of the same length.

If I make a multifasta file that has a sequence from an Multisequence alignment instead of the sequence itself, would that work for pegas and a haplotype network? Or would it then change the output? Would it even work for the conversion to a matrix?

What do people do with uneven sequence lengths?

Many thanks!

[SheepwormJM]

Ah sorry, I had totally missed the #92 issue raised by FischHa. From looking at their data I'm assuming that loading a MSA might work to get it into a matrix. So I guess my remaining question is whether it will work to then plot a haplotype network? Or will indels etc be discarded?

Thanks :)