Velvet: Retain Read Names In Afg File

Question

[Lee Katz]

This is a followup question after having fine-tuned what I really want. http://biostar.stackexchange.com/questions/2929/velvet-assembly-quality

I've started parsing the afg/ace output of Velvet and I realized that all of the reads are just numbered. For example my read with the ID "F4JLIHI02" might have the ID in the output as "113337" but there is no apparent mapping from actual read names to numbered IDs. How can I obtain this mapping, or how can I perform the assembly while retaining the read names?

My current commands are

velveth $run_name 27 -long $readsFilename
velvetg $run_name -cov_cutoff auto -exp_cov auto -read_trkg yes -amos_file yes -ins_length 2500 # I figure that the ins_length can remain there by default in case I give it a paired end read file but won't do any harm if not
amos2ace velvet_asm.afg # produces a bioperl-parsable ace file

Answer 1

[Lee Katz]

From Daniel Zerbino:

the read name <-> velvet id correspondence is stored in the header lines of the Sequences file:

grep '>' Sequences | cut -f1,2

I guess a bit of scripting would let you replace the Velvet IDs in the AFG file.

Answer 2

[Rahul Sharma]

Hi, I want to change the Id's in the velvet_asm.afg file with the read Id's in Sequence file. What Id's should I change? as there are iid, eid and rds. From where can I have the description of fields used in velvet .afg file? Best regards, Rahul

Answer 3

[SES]

The "eid" is the identifier that you could modify to reflect the sequence ID in your sequence file (see the description of the identifiers here and the same answer to this question over here). I would personally try to avoid this because it will only inflate what is probably already a huge file.