Gangolf Jobb gangolf at TREEFINDER.DE
Wed Jun 29 06:07:24 CDT 2005

German science is oppressing young scientists.

Here is another unpaid TREEFINDER:

TREEFINDER is a program to compute phylogenetic trees from molecular sequences.

The new version has more evolution models. The substitution models now
available are HKY, TN, GTR, GTR3, GTR3AG, GTR2 for nucleotides, and GTR20,
BLOSUM62, cpREV45, Dayhoff, JTT, mtMam, mtREV24, rtREV, VT, WAG for amino
acids. The heterogeneity models now include G, G+I and I (G=discrete Gamma,
I=invariable sites).

The GTR3, GTR3AG and GTR2 models have nucleotides collected in only three or
two character states to let base composition appear more stationary in some
cases. Manual re-coding of sequence files is unnecessary. The GTR20 is the
general time-reversible substitution model for amino acids and should
theoretically allow estimating all the parameters from enough data. The
collection of empirical protein models is open source and can be extended or
changed by everyone.

Sequence data can be now divided into user-defined partitions
(site-classes), assuming for each separate substitution, separate
heterogeneity, separate rate, separate composition, and separate weight. It
is even possible to have different data types in different partitions, but I
am not sure whether it is a good idea to compare or simply add the
likelihoods obtained with different numbers of character states. Any ideas?
Anyhow, I have provided a possibility to assign weights to partitions to
compensate for such effects.

Partitioning is done by including one or more so-called filters into a
sequence alignment, special named sequences consisting of digits. The
combination of all digits at a particular site determines to which partition
the site belongs. The number of partitions is practically unlimited.

"fil1" 111111222222
"fil2" 123123123123

Partitioning can be changed quickly as needed by switching on and off
filters, and one can have several groups of filters in a data set.
Partitions and combinations thereof can be extracted from a sequence
alignment using a special tool, which is accessible through a convenient
window dialog.

I have modernized the computation of edge supports, which is now based on
the SH-test with RELL approximation. More precisely, it is 100 percent minus
the maximum SH p-value over all local rearrangements around an edge. This is
more accurate than the old LRP's. Random numbers are being generated using
the fast Mersenne Twister algorithm.

The chi-square test to check stationarity has been removed because it is not
appropriate for that purpose. Sorry, I had implemented a widely used method
without thinking too much about it. Visualization of base composition,
however, is still available and was slightly extended to accommodate the new

There are some bug-fixes, and there are some more words in the manual. Among
the bugs was a numerical inaccuracy in the likelihood function that became
apparent with long sequences (e.g. > 10000 bp, depending on situation). I
recommend to recompute trees obtained from such with the new version. As
usual, the new software has been tested extensively - more extensively than
ever before. In particular, I have checked likelihood computation with up to
1100 sequences.

And now for something completely different.

Recently, somebody has offered me a half-wage student position to continue
the TREEFINDER project. Isn't this an insult for my work? Why don't all
these professors here work on half-wage? Those who failed in supporting a
promising PhD student when there was the opportunity. Those who can not make
phylogeny software. It takes much too long until a scientific career allows
raising a family because regulations were made for the professor's
convenience. Too many people agree on lousy contracts and unnecessary moves
to foreign places and are thereby lowering the income standards for all.
This situation is minimizing the efficiency of science, which is becoming a
place where academics merely wait for better opportunities. Much time and
work and public money is wasted in getting people academic degrees, and few
is spent to achieve something useful.

Professors will not tolerate a PhD student having the more interesting
project. They will blackmail him to do their stuff instead and they will
deny him a fair salary. But why wasting years on the professors project? Why
sharing one's fame and profits? Why let the professor decide where to live
and what to do? I propose that all researchers should get independent
funding for their own research, and working for others should be an
exception. Only the scientific usefulness of one's work should count, and
not one's academic degrees, age, opinion, adulation skills or the
willingness to subordinate. Requiring academic degrees is an instrument to
exploit and oppress young scientists, it supplies the professors with cheap
workers. Two decades of German over-education is enough. A science of
servants is not innovative, is not original. We need free research for all.

Being a good employee is boring. We need more adventurers, we need more
rebels. My secret is that I hardly ever did what others told me. TREEFINDER
would never have come into existence as nobody has told me to write a new
phylogeny program. I found this more useful and more interesting than what I
was supposed to do, and I did it without any compromise. Now, it is the
TREEFINDER which is my most popular project. Nobody is interested in what I
was paid for, which was just wasting my time and other people's money. As
you can see, it works all fine without the professor.

Why should I move away? Why should I change my field of research? Why am I
unemployed? - Just because this system makes me unnecessarily depend on
professors? It's a shame that nobody here in Munich is supporting my
project. It's a shame that young scientists are being oppressed in this
country. Look at them! They are sitting in my university and don't let me in.

Discontent researchers unite! Let us bring the tyranny of the professors to
its fall!

Gangolf Jobb

More information about the Taxacom mailing list