Multi-Sequence Alignment For Many Groups Of Genes

Question

[biolab]

Dear all, How to make muli-sequence alingment for many groups of genes? I make an example here.

>gene1_human
ATTTGCGTGACTGACTGC
>Gene2_human
GCGCGCATGATCCGATGACTG
>gene3_human
TGATACGATGCTGACTGACTGAC
......

>gene1_fly
ATTTGCGTGACTCTGC
>Gene2_fly
GCGCATGATCCGATGACTG
>gene3_fly
TGATACGATGCTGACTGTGAC
......

>gene1_worm
ATTTGCGTGACTCTGaC
>Gene2_worm
GCGCATGATCCGATGccACTG
>gene3_worm
TgtgGATACGATGCTGACTGTGAC
...

>gene1_mouse
ATTTGCGTGACTCTGaC
>Gene2_mouse
GCGCATGATCCGATGccACTG
>gene3_mouse
TgtgGATACGATGCTGACTGTGAC
......

I need to separately compare each gene from these species. The output likes below. They have the same length with gaps marked as - Does anyone know how to perform this analysis? Please give me some suggestions. Thank you very much!

Human   ATTTGCGTGACTGACTG-C
Mouse   ATTTGCGTGACT--CTGAC
Worm    ATTTGCGTGACT--CTGAC
Fly     ATTTGCGTGACT--CTG-C

Answer 1

[Pavel Senin]

Will clustalw work for you?

CLUSTAL 2.1 multiple sequence alignment


gene1_worm       ----ATTTGCGTGACT--CTGAC----
gene1_mouse      ----ATTTGCGTGACT--CTGAC----
gene1_fly        ----ATTTGCGTGACT--CTGC-----
gene1_human      ----ATTTGCGTGACTGACTGC-----
gene3_fly        ----TGATACGATGCTGACTG--TGAC
gene3_worm       -TGTGGATACGATGCTGACTG--TGAC
gene3_mouse      -TGTGGATACGATGCTGACTG--TGAC
gene3_human      ----TGATACGATGCTGACTGACTGAC
Gene2_human      GCGCGCATGATCCGATGACTG------
Gene2_fly        --GCGCATGATCCGATGACTG------
Gene2_worm       --GCGCATGATCCGATGCCACTG----
Gene2_mouse      --GCGCATGATCCGATGCCACTG----
                       :*..   ..*  *:

edit: yes, MUSCLE is another option, especially, it the sequences vary in their length

Gene2_human      ---GCGCGCA----TGATCCGATG--ACTG
Gene2_fly        -----GCGCA----TGATCCGATG--ACTG
Gene2_worm       -----GCGCA----TGATCCGATGCCACTG
Gene2_mouse      -----GCGCA----TGATCCGATGCCACTG
gene3_human      ---TGATACGATGCTGACTGACTG--AC--
gene3_worm       TGTGGATACGATGCTGACTG--TG--AC--
gene3_mouse      TGTGGATACGATGCTGACTG--TG--AC--
gene3_fly        ---TGATACGATGCTGACTG--TG--AC--
gene1_human      ---ATTTGCG----TGACTGACTG---C--
gene1_fly        ---ATTTGCG----TGACTC--TG---C--
gene1_worm       ---ATTTGCG----TGACTC--TG--AC--
gene1_mouse      ---ATTTGCG----TGACTC--TG--AC--
                         *     ***     **   *

[biolab]

Thanks! I just worry that I have many genes, when clustering all sequences together, some of them may not be well aligned. for example, the last C in fly and human gene1 are not aligned very well. Ideally I can run multi-sequence alignment for each gene in batch. What's your ideas? I don't know the CLUSAL algorithm.

[Pavel Senin]

clustering, as the process, is somewhat different from multiple alignment. you can pre-process you large dataset (how many sequences?) with cd-hit - which will cluster, i.e. partition, it (can handle large data) - then you can apply clustalw to clusters in order to obtain multiple alignments

[biolab]

Hi Pavel, I have ~3000 sequences. What tools for the cd-hit command? Could you please say a liitle bit more on the partition or what tools can be used for pre-process so many sequences? One of my further question is that can I run CLUSTALX(W) by command? In this way I can run it 3000 times. Thank you very much!

[Pavel Senin]

You could try to run clastalw on your set of 3K sequences as it is, I guess. CD-HIT is the software used to partition (cluster) a large dataset into groups of similar sequences. Configured by a similarity threshold and other parameters you can adjust the way it makes those groups. Assuming that sequences within a group (a cluster) are similar - clustalw will perform multiple sequence alignment for them significantly faster. Yes, you can install clustalw on your computer, and you don't need to run it 3000 times, just once.

[biolab]

Hi Pavel, I have one more question. I have just installed Clustalw. However, after typing clustalw I found a file input window pop up, then another window pop up. It's a step-by-step mode. How can I run it once by command? I can prepare 3000 gene sequence files, but don't know how to run clustalw in batch? When you have free time, could you please write to me a command for batch running clustalw (default parameters are ok)? Thank you in advance!!

[5heikki]

IMO muscle is way better than clustal..

[Pavel Senin]

sure! let's put that example too.

[5heikki]

Of course, these algorithms are designed for homologs, somehow I'm getting the idea that you're trying to align non-related sequences, which wouldn't make any sense in almost any context. More than that, if these are protein-coding genes, you should be aligning amino acids instead of nucleotides..

[biolab]

Hi 5heikki, I need to align nucleotide sequences rather than protein sequences. I am using targetscan to find miRNA targets in various species. The input file should be mutisequence alignment. Thanks!