Sequence.hs

[Vasili I. Galchin]

Hi BioHaskellers,

Why are there three Fasta representation-dependent functions (toFasta,
toFastaQual, toFastQ) doing in Sequence.hs which presumably is meant
to be Sequence representation-free code?

Vasili

[Vasili I. Galchin]

Fasta vs TwoBit ... .sorry if I am comparing apples and oranges .... I
have been away from this stuff for several months ... .

Vasili

Post by Vasili I. Galchin
Hi BioHaskellers,
Why are there three Fasta representation-dependent functions (toFasta,
toFastaQual, toFastQ) doing in Sequence.hs which presumably is meant
to be Sequence representation-free code?
Vasili

[Ketil Malde]

Post by Vasili I. Galchin
Why are there three Fasta representation-dependent functions (toFasta,
toFastaQual, toFastQ) doing in Sequence.hs which presumably is meant
to be Sequence representation-free code?

Because they are the standard formats, least common denominators, so to
speak? Also the most common inputs to tools and algorithms. If you
have a data structure that cannot easilty be converted into these, it
should probably not be an instance BioSeq (or BioSeqQual).

IIRC, TwoBit contains sequence information (and thus should be a BioSeq
instance), but no quality information (and thus no BioSeqQual instance).

Another example is Roche (and Ion Torrent)'s SFF format, which contains
sequence and quality data in addition to flowgram data.

-k

[Vasili I. Galchin]

Ketil,


I was suspicious of that ... not a pejorative comment BTW. Thx.

Vasya

Post by Ketil Malde

Post by Vasili I. Galchin
Why are there three Fasta representation-dependent functions (toFasta,
toFastaQual, toFastQ) doing in Sequence.hs which presumably is meant
to be Sequence representation-free code?

Because they are the standard formats, least common denominators, so to
speak? Also the most common inputs to tools and algorithms. If you
have a data structure that cannot easilty be converted into these, it
should probably not be an instance BioSeq (or BioSeqQual).
IIRC, TwoBit contains sequence information (and thus should be a BioSeq
instance), but no quality information (and thus no BioSeqQual instance).
Another example is Roche (and Ion Torrent)'s SFF format, which contains
sequence and quality data in addition to flowgram data.
-k
--
If I haven't seen further, it is by standing in the footprints of giants

...

[Vasili I. Galchin]

So Ketil ... if I understand your response(perhaps naively) why is
there no SFF functions exported from Sequence.hs ?? Maybe I missed
something.

Vasya

Post by Vasili I. Galchin
Ketil,
I was suspicious of that ... not a pejorative comment BTW. Thx.
Vasya

Post by Ketil Malde

Post by Vasili I. Galchin
Why are there three Fasta representation-dependent functions (toFasta,
toFastaQual, toFastQ) doing in Sequence.hs which presumably is meant
to be Sequence representation-free code?

Because they are the standard formats, least common denominators, so to
speak? Also the most common inputs to tools and algorithms. If you
have a data structure that cannot easilty be converted into these, it
should probably not be an instance BioSeq (or BioSeqQual).
IIRC, TwoBit contains sequence information (and thus should be a BioSeq
instance), but no quality information (and thus no BioSeqQual instance).
Another example is Roche (and Ion Torrent)'s SFF format, which contains
sequence and quality data in addition to flowgram data.
-k
--
If I haven't seen further, it is by standing in the footprints of giants

...

[Ketil Malde]

Post by Vasili I. Galchin
So Ketil ... if I understand your response(perhaps naively) why is
there no SFF functions exported from Sequence.hs ?? Maybe I missed
something.

Not at all - the SFF stuff is implemented in the 'biosff' library, which
imports 'biocore' and defines instances of the classes from core.

-k