Using JTT+I+G on RAxMLHPC (V. 7.2.8) CIPRES Portal
336 views
Skip to first unread message
Adam
Jan 25, 2012, 8:00:19 PM1/25/12
to raxml
Hello,
I am trying to analyze a protein alignment using the JTT+I+G model (as
determined to be the most precise by ProtTest). From the interface for
RAxML provided on the CIPRES portal, I can select JTT as the
substitution matrix. My question is: is I+G selected by choosing "Yes"
for the parameter "Estimate proportion of invariable sites (GTRGAMMA +
I)"?
If not, is there a way of running the JTT+I+G on the portal, or do I
have to download RAxML and do this manually? If so, is the "-m
PROTGAMMAImatrixName[F]" command appropriate for this situation?
Thank you,
Adam
I am trying to analyze a protein alignment using the JTT+I+G model (as
determined to be the most precise by ProtTest). From the interface for
RAxML provided on the CIPRES portal, I can select JTT as the
substitution matrix. My question is: is I+G selected by choosing "Yes"
for the parameter "Estimate proportion of invariable sites (GTRGAMMA +
I)"?
If not, is there a way of running the JTT+I+G on the portal, or do I
have to download RAxML and do this manually? If so, is the "-m
PROTGAMMAImatrixName[F]" command appropriate for this situation?
Thank you,
Adam
Adam
Jan 25, 2012, 8:07:37 PM1/25/12
to raxml
Just to add, I've read the the RAxML 7.0.4 manual, specifically the
section on p. 20 which discusses the author's personal opinion on
using I+G. So to add to my previous question, if I were to just run my
analysis with JTT, how accurate can I take the results to be,
considering that JTT alone is ranked much lower as the correct choice.
Thank you,
Adam
section on p. 20 which discusses the author's personal opinion on
using I+G. So to add to my previous question, if I were to just run my
analysis with JTT, how accurate can I take the results to be,
considering that JTT alone is ranked much lower as the correct choice.
Thank you,
Adam
Alexis
Jan 28, 2012, 8:42:44 AM1/28/12
to raxml
Hi Adam,
> I am trying to analyze a protein alignment using the JTT+I+G model (as
> determined to be the most precise by ProtTest). From the interface for
> RAxML provided on the CIPRES portal, I can select JTT as the
> substitution matrix. My question is: is I+G selected by choosing "Yes"
> for the parameter "Estimate proportion of invariable sites (GTRGAMMA +
> I)"?
That's correct.
Alexis
> I am trying to analyze a protein alignment using the JTT+I+G model (as
> determined to be the most precise by ProtTest). From the interface for
> RAxML provided on the CIPRES portal, I can select JTT as the
> substitution matrix. My question is: is I+G selected by choosing "Yes"
> for the parameter "Estimate proportion of invariable sites (GTRGAMMA +
> I)"?
Alexis
Alexis
Jan 28, 2012, 8:44:42 AM1/28/12
to raxml
Hi Adam,
> Just to add, I've read the the RAxML 7.0.4 manual, specifically the
> section on p. 20 which discusses the author's personal opinion on
> using I+G. So to add to my previous question, if I were to just run my
> analysis with JTT, how accurate can I take the results to be,
> considering that JTT alone is ranked much lower as the correct choice.
RAxML always uses a rate heterogeneity model (either CAT or GAMMA or
GAMMA+PInvar).
Hence when omitting the proportion of invariable sites option, RAxML
will still either use CAT (not to be confused with the CAT model by
Lartillot et al) or GAMMA.
My personal opinion is that by using GAMMA, you will be on the safe
side.
Alexis
> Just to add, I've read the the RAxML 7.0.4 manual, specifically the
> section on p. 20 which discusses the author's personal opinion on
> using I+G. So to add to my previous question, if I were to just run my
> analysis with JTT, how accurate can I take the results to be,
> considering that JTT alone is ranked much lower as the correct choice.
GAMMA+PInvar).
Hence when omitting the proportion of invariable sites option, RAxML
will still either use CAT (not to be confused with the CAT model by
Lartillot et al) or GAMMA.
My personal opinion is that by using GAMMA, you will be on the safe
side.
Alexis
Adam
Feb 2, 2012, 1:31:01 AM2/2/12
to raxml
Alexis,
Perfect, thank you for the helpful reply.
Adam
Perfect, thank you for the helpful reply.
Adam
Reply all
Reply to author
Forward
0 new messages