BioMoo Transcript from 30-1-98

ClareS turns the ClareS_recorder on.


ClareS says "Hi, welcome to the second meeting on Bioinformatics..."

Lizw says ""Hi, Clare and everybody""

ClareS says "Before we start properly, I should probably warn you..."

ClareS says "... that according to Christian here there was a power outage in Israel today which has affected BioMOO connectivity"

ClareS says "so it's possible that we may all lose the connection"

ClareS says "I would also like to introduce ChristianF to those of you who haven't met him yet"

ChristianF  waves "Hi, nice to meet you all!

Patricia [to ChristianF Hello] ChristianF

ClareS says "He's a consultant on the course and also on the Biocomputing course run from Bielefeld, Germany"

ClareS [to ChristianF] would you like to add anything?

Jim Pitts says "Hi"

JulieJohnston says "hi"

ChristianF [to ClareS] "only a pointer for those interested in Biology/biochemistry on the WWW

ChristianF says "Which is the Internet Biologists platform."

Jim Pitts finds his way out.

ClareS notes the URL for future reference

ChristianF  ooh "I did not mean to scare anybody ;-)

ClareS [to ChristianF] I thought it as well to mention in, just in case ;)

ChristianF  nods "Thanks for the introduction.

ClareS . o O ( in = it )

ClareS can't type

ClareS says "Well: does anyone have any questions about this section of the course?"

Jim Pitts finds his way in.

ClareS . o O ( if I don't know the answer, Christian probably will he knows more about sequence analysis than I do )

ChristianF  hides

ChristianF  smiles

ClareS grins

ClareS [to Patricia] you mailed the list about the tutorial

Patricia [to MIPS] PRINTS, aminoacid characterization window

Patricia [to Claire] Mips, PRINTS, aminoacid characterization were links that were not working

ClareS doesn't have the web available from home but will try anyway

Lizw says ""I hope I have not missed a crucial part of the Informatics material, but I haven't seen any description of how the sequence alignment algorithms work. (Not being a mathematician, maybe I don't want to know!) But there are a lot of parameter

 choices in, for example, Blast searches and one would like to have confidence that one had optimised things for one's application""

ClareS [to lizw] that is a very good point

(Patricia has disconnected.)

ClareS [to lizw] you haven't missed a crucial part because the Bioinformatics material in this course is really pretty elementary

ChristianF  nods

(Patricia has connected.)

ClareS says "Basically for bioinformatics at this level, the most important thing to realise is that you *can* trust the experts..."

Lizw says ""I am glad about that. Perhaps there is some information at the Blast sites. I haven't done much exploring.""

Pjw [guest] finds its way in.

ClareS says "you should know, more or less, how a sequence alignment works, and for any practical applications you're likely to meet at this level..."

ClareS says "it is OK just to accept the default values"

(Patricia has disconnected.)

ChristianF [to lizw] "the thing about the algorithms is that they require 'a lot of' mathematical understanding since they are based on matrices.

ClareS nods, remembering maths at degree level *a very long time ago*

ChristianF [to lizw] "if you really wanted to check out the meaning of each value and how you could possibly fine tune your search, I'd recommend checking out the Hypertext coursebook for the biocomputing course which is available online

ChristianF says "Usually, the part with the mathematics, take a great deal of time during the biocomputing course :-)"

NickyM says "oh dear"

ClareS says "yes, that's an excellent resource: but really too advanced for this course which covers sequence analysis in such a short time"

ClareS [to NickyM] what's the problem?

ChristianF says "But it is certainly helpful, since when you cannot find a match for a sequence, that does not mean that there is no match, just the parameters might be too strict"

ChristianF  nods to ClareS

(Patricia has connected.)

ClareS agrees with Christian

ClareS says "Normally you will want to test the first few matches you get out of a sequence scan to see whether they are statistically significant"

Lizw says ""Thankyou ChristianF. I am just a little concerned that I may be missing a lot of information in my searches because I get a lot of hits with one segment of sequence, upto the maximum 50 reported. I wonder if there are homologies in other par

ts of the sequence that I am just not getting to.  (The molecule I am working with is a recombinant antibody - and there are lots of antibodies in the databases!)" "

ClareS says "and for that you'd use pairwise alignment"

NickyM says "I'm working from home so its a bit slower than usual and the cat keeps jumping on the keyboard"

ChristianF  nods to LizW "If you really need the information, check out the coursebook. It will certainly get you started on this.

ChristianF  smiles to NickyM

ClareS [to NickyM] one of ours is asleep in the box I put paper to recycle in ;)

Jim Pitts says "It must be after the mouse!"

ClareS grins

ClareS [to lizw] antibodies are a difficult case

Lizw says "" Thankyou, ChristianF. I will do that.""

ClareS [to lizw] as there are conserved and variable regions...

JulieJohnston says "That was useful because I too work with antibodies"

Lizw says ""Thanks ClareS, for that encouragement!""

ChristianF  laughs

ClareS [to lizw] the sequence alignments will tend to just home in on antibody sequences by matching the conserved regions

Patricia [to lizw] Have you done all of your sequence comparisons on the same software? Perhaps, compare it with another software, another computer.

ClareS [to ChristianF] do you know much about the Kabat antibody database?

(Pjw [guest] has disconnected.)

ChristianF [to ClareS] "No, not much. I haven't used it

NickyM says "I'm glad that the topic of large numbers of hits is being discussed. Ican only work with the examples in section 5 at the moment but can't work out how you decide which hit is the most 'correct' if it isn't necessarily the first in the list.


Lizw says ""I have done most of it on WU-Blast. Is there anything that might suit this application better?""

ClareS [to ChristianF] I am thinking that if one were to search a specialist database of antibody sequences, possibly with software optimised to handle sequences with conserved and variable regions, it would help...

Patricia [to LizW,] I was wondering too. Perhaps ClareS or ChristianF can tell us

ClareS [to ChristianF] but I am not an antibody expert and I don't know if such a facility is available

ChristianF [to NickyM] "could you give me the URL of the page you're referring to?

ChristianF [to ClareS] "let me search for an URL

JulieJohnston says "I was using the Kabat data base today and you could only search using nucelotide sequence"

ClareS really *must* get the web set up at home

NickyM says "slowly getting there"

Patricia [to ClareS] Are there several versions of Blast?What are the differences?

ClareS [to JulieJohnston] yes...but if you *have* the nucleotide informaion available it would be as well to use it

ClareS [to Patricia] there are many different Blast *servers*, which obviously isn't what you were asking...

ChristianF [to Patricia] "Differences between BLAST versions are partly due to optimizations of the code

ChristianF says "ie. the search algorithms"

ClareS [to Patricia] but they may well use different *matrices* or give different *default parameters*

Patricia [to ClaireS] From the point of view of algorithms, I wonder which is the "best" from a biologist's or chemist's point of view.

ClareS [to Patricia] so they may not give *exactly* the same result

ChristianF [to lizw] "have you done a FASTA search as well?

JulieJohnston says "I guess it depends on your application. I'm generally looking for sequences that are similar at the protein sequence for modelling purposes and to see what they are antibodies to."

ClareS says "Yes, that's a good point: if you get ambiguous results with Blast, you should try with FastA and vice versa"

Lizw says ""Yes, I started with Fasta. The results are almost identical with what I get out of Blast.""

ChristianF  nods

ClareS [to lizw] have you tried with different matrices?

Lizw says ""Do you mean using different databases or different sections of the sequence?""

ClareS [to JulieJohnston] I am sure there must be a specialist database of antibody sequences somewhere, at the protein level, but I don't know of it... if Kabat doesn't give this information

Patricia [to ChristianF] Don't you mean different "Dayhoff" matrices?

ChristianF [to Julie&LizW] "You might wanna check out the http

ClareS [to lizw] neither

ChristianF says "they have links to ab databases"

Lizw says ""What's a Dayhoff matrix?""

ClareS [to ChristianF] I have a carriage return in the wrong place

ChristianF [to Patricia] "I'm not exactly sure about the modifications that were applied to the different BLAST or FastA versions, but they can include the use of different matrices or modifications of the search algorithms itself.

Lizw says ""No I haven't messed about with the default settings yet, cos I don't know what their significance is!""

ChristianF [to ClareS] "No, it's one word

Lizw says ""Thanks for that URL.""

Patricia [to ChristianF] Are different matrices available on the WEB?

ChristianF [to lizw] "Try http

ChristianF [to Patricia] "Yes, a very comprehensive overview is at http

ClareS says "again speaking from a biologist's point of view, so *not mathematically*, you might use an "identity" matrix, so score 1 for an exact match of amino acids, of you might have different scores for identity and for similarity (F,Y; I.L.V; D,E o

r whatever)"

ChristianF [to Patricia] "there are also help files available on the individual BLAST sites which specify the parameters used.

Patricia [to lizw] May I suggest a textbook...."Introduction to Computational Molecular Biology" by setubal and Meidanis, PWS Publishing Co, 1997

Lizw says "" As my ultimate objective is 3D modeling, I have been concentrating on the PBD databse.""

ClareS is thinking of *writing* a textbook, fairly soon

ClareS . o O ( at *very* introductory level )

Patricia [to ClareS] You're thinking of writing a book...great!

ClareS [to Patricia] only thinking, so faar

Lizw says ""Thanks Patricia, I think I could really use a real, rather than virtual, textbook at this stage.""

ChristianF [to lizw] "Try that server in France

Patricia [to ClareS] A book for beginners is needed, a first book.

ChristianF [to lizw] "the virtual text book is available as postscript as well, so you could just print it out

ChristianF  nods to ClareS&Patricia

ClareS [to Patricia] I would like to reinforce what Christian said about help files on Blast sites. They can be very useful and *fairly* user friendly

Patricia [to ClareS] OK But are they all using the same matrices?

ClareS [to Patricia] there are a variety of widely used matrices. most of these will be available at most Blast sites (for example)...

ClareS [to Patricia] ... but they may not all use the same matrix as the *default* (ie the one you get if you don't select one explicitly ;)

ClareS [to ChristianF] do you have a "favourite" Blast site?

ChristianF [to Patricia] "if you check out http

Patricia [to ClareS] It would be a bit of work to look into the derivation of the matrices...I wonder ..there must be a review of them somewhere.

ChristianF [to ClareS] "that's the one ;-)

ClareS nods

ClareS says "before we go any further, I would like to warn you that I will be closing the"formal" part of this meeting at midnight sharp!"

ClareS says "it is getting very late in the UK and my phone bill is also growing fast"

ChristianF [to Patricia] "http

Lizw says ""But ClareS, the day is just beginning here...""

Patricia [to ChristianF] OK. The original literature. Thanks for the URL

ClareS [to ChristianF] meaning the original papers rather than reviews or textbooks, I suppose?

ClareS [to lizw] good morning ;)

ChristianF [to Patricia] "There is not much info available online and that's one of the reasons the biocomputing course was developed in 1994/95

ClareS wonders if Christian will have a few recruits for his course amongst PPS students ;)

Patricia [to ChristianF] I haven't used Blast in 1 1/2 years since a course at Brandeis U. Our conversation reminds me about the matrices and parameters.

ChristianF [to ClareS] "the biocomputing course is not offered by me. There might be another edition of the course during the summer which will probably be headed by Rebecca Parsons from U Florida.

ClareS nods at christian and blushes

Patricia [to ChristianF] Would that be the VSNS course?

ChristianF says "Those of you who'd be interested in participating should subscribe themselves to the bcd-prospective mailing lists and then they will get any relevant announcements"

ChristianF  nods to Patricia

NickyM says "I'm afraid this talk of matrices is over my head, does this mean I've missed something important - i'm an electron microscopist not molecular biologist"

Patricia [to ChristianF] How to subscribe to list?

ClareS [to NickyM] you really don't need to worry about it, not for the PPS course, at least

ClareS [to NickyM] you should worry more about the *biological implications* of sequence homology

ChristianF [to Patricia] "e-mail to with subscribe vsns-bcd-prospective

ChristianF [to Patricia] "Or you can also e-mail Georg Fuellen or page him here in BioMOO. He's usually here during the day

DavidM finds his way in.

Jim Pitts says "Hi David"

Patricia [to ChristianF] OK

ChristianF  nods to NickyM "You'd only have to worry about matrices when you do extensive sequence alignment

ClareS [to DavidM] Hi - we have been having a very interesting discussion

DavidM says "Hi there!"

ClareS is thinking of turning the recorder off soon

Jim Pitts waves

DavidM says "Have you had a good session?"

ClareS leaves that to the students to answer, but thinks so ;)

ChristianF says "any other questions re. bioinformatics?"

ClareS says "Georg Fuellen (GeorgF in BioMOO) is a very useful person to get to "know" - very friendly, very knowledgeable about bioinformatics, and very often in BioMOO!"

ChristianF  nods

Patricia [to Oh] we'll have to do that.....

ClareS says "Well, if there are no more questions or problems, I think it must be time for bed"

ClareS says "thank you all, and especially Christian, for contributing to such a good discussion"

Lizw says "" I think I'll be off now. It has been a good session. Thanks everyone for your help. Goodnight ClareS.""

ClareS turns the ClareS_recorder off.