Figure 1: I was confused how the two tress are compared by fit on the upper panel. And on the lower panel I wanted to know what the solid line meant and why is the position of cofilin so low.
Figure 2: I wanted to know what the x-axis and y-axis meant and what the red arrow signifies. I do understand what the information score means but not the relation with frequency and also not the red arrow.
The audience are not bioinformaticist, and neither am I. Therefore a simple and as less technical as possible explanation would really help. I would appreciate if there is any paper or book chapter or ppt that could help me.
I don't see much relevance to bioinformatics here, but perhaps others see it differently, so I will not close the question.
What have you tried here: Have you read the paper? Just looking at the two figures it's hard to answer your question succinctly after a brief look and I don't feel like I have the time to download and read the paper just to answer your question here. It's not that I wouldn't like to help out here, it's just I don't know if I have to time to commit to this when I am not certain you have committed much time to understanding the figures on your own.
Understanding research is difficult: One of my early mentors told me to "just keep reading" and encouraged me to not be frustrated when I didn't understand something. If you don't understand a paper, read it again. Read it again after that. Go back to other citations and read those. Then go back and read the original paper again. If you are going to survive in science you'll have to be able to comprehend research independently without the help of others (or at least with minimal help). It's not an easy thing to learn, but it's essential that you do so. Trust me in that it gets easier and proceeds much faster the more you do it. Best of luck to you!
+1 This is a very good response, but I think the question should be closed. It's not related to bioinformatics, and it's clear that the OP just wants a paper explained. I think it is okay to ask for an explanation if posed in the right way, but it does not seem there is actual question here that is appropriate for this site.
It is ok to state that the question is too difficult to answer and perhaps overly generic. But that should not be grounds for closing a question after all this is about interpreting a QTL plot and phylogenetic trees. Both of these are valid bioinformatics topics.
What is most important throughout is to refrain from making and stating judgmental calls of what the OP wants or does not want. No question should ever be closed because you think it is a homework problem, or because you think original poster should put in more effort, or because the OP just wants a paper explained. If you disagree with what your perceive to be the underlying motivation for the post you can just simply to ignore the question and let someone else step in if so they wish.
When you close a question you are also precluding others from answering. This is only be acceptable if the answer for the question would NOT be suitable for the topic of the site.
Fair enough, but it looks like others agreed that it should be closed and there are usually multiple opinions on these decisions. My concern is that people will lose interest in the site if the content devolves to the basic "explain this" type of homework question.
I think my posting of the paper gave a different meaning. I posted the paper to back up my question, if my question was not clear enough then anyone interested could refer the paper. Also, because I was not able to post figures here. All I wanted to understand was the difference between a data that has undergone merge analysis and one that has not, and how merge analysis works in a bigger picture context because I did read papers that gives the statistical details and equations, and I was still not able to understand the big picture which I wanted to explain my audience by interpreting the figures from the paper. Also, for the second portion, I do understand what information score means and how enrichment is scored. Here, I am having problem understanding what the peaks of frequency means and how frequency is involved in analyzing data. If the high frequency in figure 2a means something compared to lower frequency in figure 2b.
Thank you for clarifying your question. Hopefully, you didn't take the comments the wrong way because we want to help but it was not completely clear before what you had tried/wanted. I'll take a look at the paper for you....
Essentially they are using this 'merge analysis' approach to refine their QTLs down to candidate genes contributing to a phenotype by incorporating the genetic relatedness of the strains. They then use a statistical model to compare the results of the merge analysis (logP statistic) with the statistic from the unconstrained model to infer whether a variant could be the QTL allele (that is, if the statistic from the merged analysis exceeds that of the unmerged/unconstrained analysis). What you see in Fig. 1 is a demonstration of this method. Fig. 2 is also demonstrating something novel, that which they call the information score. It seems this is really just showing the association of GO terms (functional annotations) with the genomic intervals identified by the merge analysis, and they compare this with a random sample of genes to see how well each set of genes is grouping.
They started with a large number of relatively large QTLs from previous studies, and this approach allowed them to narrow the list and more importantly, identify a much smaller genomic region contributing to the phenotypes of interest.
Essentially they are using this 'merge analysis' approach to refine their QTLs down to candidate genes contributing to a phenotype by incorporating the genetic relatedness of the strains. They are trying to solve the age-old issue of any QTL analysis, and that is how to go from a QTL (many, in this case) for a phenotype down to the gene, and ultimately the underlying variant. In this study, they have the benefit of knowing the ancestry of the strains, and so they can use this to drill down to smaller/fewer QTL. Their explanation of the approach is pretty straightforward, "Since a QTL must lie in a region where sequence differences distinguish strains in the same way as the QTL alleles, in a cross between two strains the QTL must lie in a region that is not identical by descent." So, they can use this "mosaic" pattern in the genome to dissect the phenotypic response.
They use a statistical model to compare the results of the merge analysis (they call it a logP statistic) with the statistic from the unconstrained model to infer whether a variant could be the QTL allele (that is, if the statistic from the merged analysis exceeds that of the unmerged/unconstrained analysis). What you see in Fig. 1 is a demonstration of this method with the model results in the top part of the figure and each one of the QTL and associated logP statistics in the bottom part. Fig. 2 is also demonstrating something novel, that which they call the information score. It seems this is really just showing the association of GO terms (functional annotations) with the genomic intervals identified by the merge analysis, and they compare this with a random sample of genes to see how well each set of genes is grouping. The part B in Fig. 2 is showing that no significant results can be found if all genes in each QTL are included, and the merge analysis is ignored. There are some specifics genes they focus on, but basically they do an enrichment test with the results from the merge analysis to identify the GO terms. I can add that the group of genes they are describing are fundamental components in plant cytoskeletons, as well.
I don't see much relevance to bioinformatics here, but perhaps others see it differently, so I will not close the question.
What have you tried here: Have you read the paper? Just looking at the two figures it's hard to answer your question succinctly after a brief look and I don't feel like I have the time to download and read the paper just to answer your question here. It's not that I wouldn't like to help out here, it's just I don't know if I have to time to commit to this when I am not certain you have committed much time to understanding the figures on your own.
Understanding research is difficult: One of my early mentors told me to "just keep reading" and encouraged me to not be frustrated when I didn't understand something. If you don't understand a paper, read it again. Read it again after that. Go back to other citations and read those. Then go back and read the original paper again. If you are going to survive in science you'll have to be able to comprehend research independently without the help of others (or at least with minimal help). It's not an easy thing to learn, but it's essential that you do so. Trust me in that it gets easier and proceeds much faster the more you do it. Best of luck to you!
+1 This is a very good response, but I think the question should be closed. It's not related to bioinformatics, and it's clear that the OP just wants a paper explained. I think it is okay to ask for an explanation if posed in the right way, but it does not seem there is actual question here that is appropriate for this site.
It is ok to state that the question is too difficult to answer and perhaps overly generic. But that should not be grounds for closing a question after all this is about interpreting a QTL plot and phylogenetic trees. Both of these are valid bioinformatics topics.
What is most important throughout is to refrain from making and stating judgmental calls of what the OP wants or does not want. No question should ever be closed because you think it is a homework problem, or because you think original poster should put in more effort, or because the OP just wants a paper explained. If you disagree with what your perceive to be the underlying motivation for the post you can just simply to ignore the question and let someone else step in if so they wish.
When you close a question you are also precluding others from answering. This is only be acceptable if the answer for the question would NOT be suitable for the topic of the site.
Fair enough, but it looks like others agreed that it should be closed and there are usually multiple opinions on these decisions. My concern is that people will lose interest in the site if the content devolves to the basic "explain this" type of homework question.
Thank you for the responses.
I think my posting of the paper gave a different meaning. I posted the paper to back up my question, if my question was not clear enough then anyone interested could refer the paper. Also, because I was not able to post figures here. All I wanted to understand was the difference between a data that has undergone merge analysis and one that has not, and how merge analysis works in a bigger picture context because I did read papers that gives the statistical details and equations, and I was still not able to understand the big picture which I wanted to explain my audience by interpreting the figures from the paper. Also, for the second portion, I do understand what information score means and how enrichment is scored. Here, I am having problem understanding what the peaks of frequency means and how frequency is involved in analyzing data. If the high frequency in figure 2a means something compared to lower frequency in figure 2b.
Thank you
Thank you for clarifying your question. Hopefully, you didn't take the comments the wrong way because we want to help but it was not completely clear before what you had tried/wanted. I'll take a look at the paper for you....
Essentially they are using this 'merge analysis' approach to refine their QTLs down to candidate genes contributing to a phenotype by incorporating the genetic relatedness of the strains. They then use a statistical model to compare the results of the merge analysis (logP statistic) with the statistic from the unconstrained model to infer whether a variant could be the QTL allele (that is, if the statistic from the merged analysis exceeds that of the unmerged/unconstrained analysis). What you see in Fig. 1 is a demonstration of this method. Fig. 2 is also demonstrating something novel, that which they call the information score. It seems this is really just showing the association of GO terms (functional annotations) with the genomic intervals identified by the merge analysis, and they compare this with a random sample of genes to see how well each set of genes is grouping.
would you mind adding this as an answer as well - just trying to get improve our "answered questions" stats
I added/expanded my comment to make it more of an answer.
Thank you SES !
They started with a large number of relatively large QTLs from previous studies, and this approach allowed them to narrow the list and more importantly, identify a much smaller genomic region contributing to the phenotypes of interest.