y*****s
发帖数: 87 | 1 http://liorpachter.wordpress.com/
Let's see what will MIT have to do on this. |
P****d
发帖数: 564 | 2 He was such a wunderkind.
I doubt that there'll be any formal investigation from MIT.
【在 y*****s 的大作中提到】
: http://liorpachter.wordpress.com/
: Let's see what will MIT have to do on this.
|
l****m
发帖数: 751 | 3 都是领域的大人物啊。
Lior Pachter的质疑应该是准确的。
跟Manolis Kellis meeting中见过好多次，感觉这个人有点不踏实，太偏biology，而
biology整个领域都浮躁喜欢钻空子。
不知道Eric lander和Cole Trapnell有什么反应。
一定会非常有趣。楼主帮我们更新啊。 |
s*********x
发帖数: 1923 | 4 连文章都不用看，我是肯定相信Lior Pachtor的。 |
l****m
发帖数: 751 | 5 Lior Pachter 对 Albert-László Barabási 也是这么不给面子啊。
牛人真是有脾气。
The network nonsense of Albert-László Barabási
http://liorpachter.wordpress.com/2014/02/10/the-network-nonsens |
s******a
发帖数: 472 | 6 very interesting!
【在 y*****s 的大作中提到】
: http://liorpachter.wordpress.com/
: Let's see what will MIT have to do on this.
|
y*****s
发帖数: 87 | 7 I think Lior is extraordinary mostly in the sense that he is not afraid of
speaking out. There are so many nonsense in this field and most people seems
okay with it. He is boycotting the RECOMB after he is convinced that this
meeting is more about politics than science. Like Lior said: Kellis is not
one of those scientists but is playing leading role in modENCODE, ENCODE,
and is serving committee roles in NHGRI, et al.
He has been spending so much time on this if you read back his last two
posts "why I read those nonsense network papers", all the comments, back and
forth arguing between him and Kellis, and people from Kellis group. It is
clear Kellis is fraudulent and has been fraudulent since 8 years ago at the
least. I do believe this two posts from Lior have gained enough attention.
Again! We will see what MIT has to do on this. |
M*P
发帖数: 6456 | 8 Manolis has no background in networks.
It is likely that he is over trusting his postdoc on this matter.
【在 y*****s 的大作中提到】
: http://liorpachter.wordpress.com/
: Let's see what will MIT have to do on this.
|
M*P
发帖数: 6456 | 9 just checked lior's post on Barabasi.
my conclusion is: he is just a weaning baby.
BTW, his cufflinks sucks too.
【在 y*****s 的大作中提到】
: http://liorpachter.wordpress.com/
: Let's see what will MIT have to do on this.
|
l****m
发帖数: 751 | 10 I would rather say "lior is an old-fashioned scientist."
While "Barabasi is a politician"
【在 M*P 的大作中提到】
: just checked lior's post on Barabasi.
: my conclusion is: he is just a weaning baby.
: BTW, his cufflinks sucks too.
|
|
|
M*P
发帖数: 6456 | 11 晕啊，说kellis的GO analysis是cherrypicking。他的cufflinks文章就不是
cherrypicking? 不也是从几百上千个显著的alternative splicing里面挑了两三个来
验证？
★ 发自iPhone App: ChineseWeb 7.8
【在 l****m 的大作中提到】
: I would rather say "lior is an old-fashioned scientist."
: While "Barabasi is a politician"
|
l****m
发帖数: 751 | 12 你在说哪篇文章？
Cufflinks2010是有湿实验验证的.
2010年RNAseq的文章，有软件，有漂亮的统计模型，有seq的数据，有湿实验验证，发
Nature biotechnology 有问题吗？即使是无名学校PI的文章当时让我来审也应该是过。
反过来，Albert-László Barabási 文章如果是无名小卒写的能上nature吗？
【在 M*P 的大作中提到】
: 晕啊，说kellis的GO analysis是cherrypicking。他的cufflinks文章就不是
: cherrypicking? 不也是从几百上千个显著的alternative splicing里面挑了两三个来
: 验证？
:
: ★ 发自iPhone App: ChineseWeb 7.8
|
M*P
发帖数: 6456 | 13 you can count how many transcripts they validate in that paper with one
hand.
and the assembly method is total bullshit.
I believe GO better, because it is at least hard to cherrypick genes to make
an GOTerm enrichment.
biotechnology
【在 l****m 的大作中提到】
: 你在说哪篇文章？
: Cufflinks2010是有湿实验验证的.
: 2010年RNAseq的文章，有软件，有漂亮的统计模型，有seq的数据，有湿实验验证，发
: Nature biotechnology 有问题吗？即使是无名学校PI的文章当时让我来审也应该是过。
: 反过来，Albert-László Barabási 文章如果是无名小卒写的能上nature吗？
|
l****m
发帖数: 751 | 14 wet-experiment validation of a handful genes is the standard way in biology
research, just pick up any nature or science paper.
I would definitely prefer "Wet-experiment validation of a single gene" than
"GO validation". The former can easily boost your paper impact factor by 3.
I happen to developed a tool for GO. It is not as robust as it seems.
The work in 2010 Cufflinks paper is enough for nature biotechonology. It is
not a pure computation paper anyway.
make
【在 M*P 的大作中提到】
: you can count how many transcripts they validate in that paper with one
: hand.
: and the assembly method is total bullshit.
: I believe GO better, because it is at least hard to cherrypick genes to make
: an GOTerm enrichment.
:
: biotechnology
|
M*P
发帖数: 6456 | 15 I suspect that the problem of robustness of GO analysis comes from the test
everyone is using.
can you share the paper with me?
my point is that validating one or two genes from thousands of genes
predicted is no different from pick one or two go terms from a list of
enriched term.
biology
than
.
is
【在 l****m 的大作中提到】
: wet-experiment validation of a handful genes is the standard way in biology
: research, just pick up any nature or science paper.
: I would definitely prefer "Wet-experiment validation of a single gene" than
: "GO validation". The former can easily boost your paper impact factor by 3.
: I happen to developed a tool for GO. It is not as robust as it seems.
: The work in 2010 Cufflinks paper is enough for nature biotechonology. It is
: not a pure computation paper anyway.
:
: make
|
M*P
发帖数: 6456 | 16 how about the second cufflinks paper? you believe the figure 1 BS there?
how many validation did they do on the second paper?
lior is no different from Barabasi
laughing at Barabasi'a linear algebra does not prove lior saw the big
picture problem first.
过。
【在 l****m 的大作中提到】
: 你在说哪篇文章？
: Cufflinks2010是有湿实验验证的.
: 2010年RNAseq的文章，有软件，有漂亮的统计模型，有seq的数据，有湿实验验证，发
: Nature biotechnology 有问题吗？即使是无名学校PI的文章当时让我来审也应该是过。
: 反过来，Albert-László Barabási 文章如果是无名小卒写的能上nature吗？
|
l****m
发帖数: 751 | 17 " I suspect that the problem of robustness of GO analysis comes from the
test
everyone is using.
can you share the paper with me? "
Just pick up 1000 genes by random, and do a GO analysis, you will understand
what I meant. GO always returns lots of result, leaving the authors plenty
of space to pick whatever he wants.
" my point is that validating one or two genes from thousands of genes
predicted is no different from pick one or two go terms from a list of
enriched term. "
There is actually a huge difference in cost. The cost in wet-experiment
prohibit the number of experiments you can do, and to some degree prevent
selective results.
test
【在 M*P 的大作中提到】
: I suspect that the problem of robustness of GO analysis comes from the test
: everyone is using.
: can you share the paper with me?
: my point is that validating one or two genes from thousands of genes
: predicted is no different from pick one or two go terms from a list of
: enriched term.
:
: biology
: than
: .
|
A*****n
发帖数: 243 | 18 问题是Manolis Kellis最后用来说故事的两个function term在GO analysis里面连
top10或者top20都算不上啊，这个就有点问题了。
"In fact, retinal cone cell development and nerve growth factor were 33 and
34 out of the 55 listed GO categories when sorted by the DAF p-value (42 and
54 when sorted by heterozygosity p-value)"
【在 M*P 的大作中提到】
: 晕啊，说kellis的GO analysis是cherrypicking。他的cufflinks文章就不是
: cherrypicking? 不也是从几百上千个显著的alternative splicing里面挑了两三个来
: 验证？
:
: ★ 发自iPhone App: ChineseWeb 7.8
|
l****m
发帖数: 751 | 19 " how about the second cufflinks paper? you believe the figure 1 BS there? "
Which one you are talking about?
" lior is no different from Barabasi"
Your previous posts show only inadequate experience in bioinformatics, which
is totally fine. Nobody is expert in all field.
However, this single one suggests who you are. You don't really care what is
right. Barabasi apparently published too many "big picture problem" in the
form of "original research article". If his paper is "perspective/editorial/
communication", then nobody (including lior) will question him. |
r****t
发帖数: 10904 | 20 没什么意思啊,这种事情还吵. nature biotech 审稿估计水平不够没看出问题来.
lior 受气很久了, 领域里面假科学混得好的不少, 打假斗士出现应该受欢迎才对.
【在 y*****s 的大作中提到】
: http://liorpachter.wordpress.com/
: Let's see what will MIT have to do on this.
|
r****t
发帖数: 10904 | 21 我X, Feizi et al. 的审稿人估计 college 线性代数都不过关,
nature 选审稿人的方式有问题.
edit: 私自改图这个事情,MIT 应该处理了
【在 r****t 的大作中提到】
: 没什么意思啊,这种事情还吵. nature biotech 审稿估计水平不够没看出问题来.
: lior 受气很久了, 领域里面假科学混得好的不少, 打假斗士出现应该受欢迎才对.
|
