Reg. BLAST

[jimmm....]

Hello all,

when we do BLAST, How to get the homology between sequence. I mean to
say on which value we need to decide the homology. On E-value or on
Score. which value does tell the homology.

With reg
Sumeeth Sabade

[elu...@fca.unl.edu.ar]

Hi Sumeeth Sabade
Homology between sequences only can be deduced performing a
phylogenetic analisys. What you obtain from BLAST is sequence
"similarity".
Best
--
Lic. Leandro Lucero
Catedra Morfología Vegetal
Facultad de Ciencias Agrarias
UNIVERSIDAD NACIONAL DEL LITORAL
Kreder 2805 - S3080HOF - Esperanza - Prov. de Santa Fe - Argentina

Mensaje escrito por: "jimmm...." <sumith...@gmail.com>:

> --
> You received this message because you are subscribed to the Google
> Groups "PredictProtein-Science" group.
> To post to this group, send email to
> predictprot...@googlegroups.com
> To unsubscribe from this group, visit
> http://groups.google.com/group/predictprotein-science/subscribe?hl=en_US
> For more options, visit this group at
> http://groups.google.com/group/predictprotein-science?hl=en
>
> This group is affiliated with the PredictProtein server -
> http://www.predictprotein.org/.
> PredictProtein is an Internet service for sequence analysis and the
> prediction of protein structure and function from sequence.
> To get PredictProtein news and updates follow us on twitter:
> http://twitter.com/#!/rostlab
>

[karim mezhoud]

Hello,
When we use local blast: similarity > 80%
             E-value < 10 exp(-5)

based on your context you can adapt the alignment by chosing BLOSUM matrix: BLOSUM80 for nearly species
    BLOSUM40 for far species

regards

--
You received this message because you are subscribed to the Google
Groups "PredictProtein-Science" group.
To post to this group, send email to
predictprot...@googlegroups.com
To unsubscribe from this group, visit
http://groups.google.com/group/predictprotein-science/subscribe?hl=en_US
For more options, visit this group at
http://groups.google.com/group/predictprotein-science?hl=en

This group is affiliated with the PredictProtein server - http://www.predictprotein.org/.
PredictProtein is an Internet service for sequence analysis and the prediction of protein structure and function from sequence.
To get PredictProtein news and updates follow us on twitter:
http://twitter.com/#!/rostlab

--
Karim Mezhoud, PhD. ir.
Molecular Biology Lab
National Center of Nuclear Sciences and  Technologies
Technopole Sidi Thabet, 2020, Ariana, Tunisia
https://wiki.edubuntu.org/kmezhoud/Bioinformatics
FRE 3206 CNRS-USM 0505 Molécules de communication et adaptation des micro-organismes,
Museum national d'Histoire naturelle, 12 rue Buffon, F-75231 Paris cedex 05, France.

[Suhaila Sulaiman]

Hi,

In my opinion, we cannot rely on E-value only, because E-value is dependent on the database size. Let say there are many sequences which actually not that correct in the database, so it affects the database size, thus affecting the E-value as well.

Actually in BLAST the best way to know the significant hit is by observing the pairwise alignment. We can check for the query length vs match length to take into account because BLAST is a local alignment tool, so sometime 10-residue sequence can be 100% identical to hit sequence even the query sequence is 1000 residue!

Always look at various parameter at one time: percentage identity, score, query and match length, E-value.

Regards,

--
You received this message because you are subscribed to the Google
Groups "PredictProtein-Science" group.
To post to this group, send email to
predictprot...@googlegroups.com
To unsubscribe from this group, visit
http://groups.google.com/group/predictprotein-science/subscribe?hl=en_US
For more options, visit this group at
http://groups.google.com/group/predictprotein-science?hl=en

This group is affiliated with the PredictProtein server - http://www.predictprotein.org/.
PredictProtein is an Internet service for sequence analysis and the prediction of protein structure and function from sequence.
To get PredictProtein news and updates follow us on twitter:
http://twitter.com/#!/rostlab

--
Suhaila Sulaiman

[Saurabh Raghuvanshi]

Hi,
A lot depends on what you want to infer from the alignment.
1. E-value can be used to filter out significant hits form junk alignments. Any thing below 10 exp( -10) is a good hit. If the sequence origin is much different one can use < 10 exp(-5). Such parameters have been used to define functional similarities between sequences in most studies.

2. Further one can look at the coverage of alignment ie what length of your sequence is involved the alignment.

Again answer would be more precise if your objective is clear.

Best

Saurabh

[sneha paul]

we decide based on the e - value. u c the red bands .. the most similar sequence will b next to the qury sequnce .. and the res color band indicates the the max similarity

--
You received this message because you are subscribed to the Google
Groups "PredictProtein-Science" group.
To post to this group, send email to
predictprot...@googlegroups.com
To unsubscribe from this group, visit
http://groups.google.com/group/predictprotein-science/subscribe?hl=en_US
For more options, visit this group at
http://groups.google.com/group/predictprotein-science?hl=en

This group is affiliated with the PredictProtein server - http://www.predictprotein.org/.
PredictProtein is an Internet service for sequence analysis and the prediction of protein structure and function from sequence.
To get PredictProtein news and updates follow us on twitter:
http://twitter.com/#!/rostlab

--
Regards
Sneha

JMJ 

OSDD

[Teh Ban Hong]

Hi,

After BLAST is carried out, I think homology can be deduced from sequence identity (which is the most important one) and the E-value is considered as an indicator to tell how reliable the sequence obtained is similar to our query. Sequence identity is also one of the result figure obtained from pairwise sequence alignment. Sequence similarity is taken into consideration also. Also, we need to observe the sequence length and alignment length.

While for phylogenetic analysis, it is one of the deduction only can be done after multiple sequence alignment. I think it is another way to tell about the homology. Phylogenetic analysis can tell about the relationship among the sample of study, but all depends on the results obtained from the multiple sequence alignment. 

Hope it helps. Please correct me if anything goes wrong for my explanation here.

---

From: Saurabh Raghuvanshi <sau...@genomeindia.org>
To: predictprot...@googlegroups.com
Sent: Wednesday, 15 December 2010 01:52:31
Subject: Re: [predictprotein-science:503] Reg. BLAST

[STEPHEN H. PILDER]

e-value is not a measure of homology (whether  genes are related by descent). It is a measure of similarity, however.