Dear all,
I have a dumb question that bothers me a lot. I have a set of genes with calculated ChIP-seq binding intensities. Notably, most of these genes have more than one value (i.e., because of multiple binding I thought). Now I want to clean these data by calculating the average of biding for each gene. The head of the data likes the following,
A2M -3.63
A2M -18.46
ACE 1.68
ACYP2 8.23
ACYP2 6.09
ADAMTS12 1.77
ADAMTS12 1.65
ADAMTS6 5.36
ADIPOQ 4.06
ADIPOQ 1.71
AGR2 13.71
AGR2 8.92
AGR2 5.05
AGR2 3.02
AGR2 2.23
AGR2 -4.22
I have these data in excel file. Obviously, it is annoying to do manually in Excel because of thousands of genes. How do I do this? I guess a simple script may help. But I have no experience in programming. Your help is much appreciated! Thank you! Xiaoyong Fu Houston, TX
•
written
by
xiaoyonf •
Assuming the tab delimited file:
sort -k1,1 in.txt | groupBy -g 1 -c 2 -o mean > out.txt
out.txt:
A2M -11.045
ACE 1.68
ACYP2 7.16
ADAMTS12 1.71
ADAMTS6 5.36
ADIPOQ 2.885
AGR2 4.785
groupBy from bedtools
Hi Goutham, This looks so cool! I will try it definitely. Thank you so much! Xiaoyong
The bedtools groupby command works beautifully to perform such work. Thank you for quick answer! Xiaoyong
You can also do this easily in R.
# reading in the data; insert your data import method here:
gene_stats <- "gene\tvalue
A2M\t-3.63
A2M\t-18.46
ACE\t1.68
ACYP2\t8.23
ACYP2\t6.09
ADAMTS12\t1.77
ADAMTS12\t1.65
ADAMTS6\t5.36
ADIPOQ\t4.06
ADIPOQ\t1.71
AGR2\t13.71
AGR2\t8.92
AGR2\t5.05
AGR2\t3.02
AGR2\t2.23
AGR2\t-4.22
"
gene_stats_df_df <- read.table(textConnection(gene_stats), header = TRUE, sep = '\t')
# get the mean for each gene in the table
aggregate(value ~ gene, data = gene_stats_df_df, FUN = mean)
output:
gene value
1 A2M -11.045
2 ACE 1.680
3 ACYP2 7.160
4 ADAMTS12 1.710
5 ADAMTS6 5.360
6 ADIPOQ 2.885
7 AGR2 4.785
You can also copy the first column (A) to another column(D), then apply remove duplicates to this column.
Then, in the E1 cell, you only have to use =AVERAGEIF(A:A, D1, B:B). With double click on the lower right corner of the E1 cell, you can copy this function for all the E column.
Thank you, aquaq. This is I thought the quickest way to solve my problem!
In Excel, select the table, go to Inset, PivotTable, choose the location (Existing Worksheet and select a blank cell). A side bar section PivotTable Fiedls will appear. Drag the Gene field name to Rows and Value to the values. The default will be the sum of values for each gene. To change it click on the drop down menu and then value field settings and choose average
(suggesting only because Pierre mentioned this as the option :P)
X <- aggregate ( x, data, mean)
Log in to answer this question.
You say some of your genes have multiple values, but your example data doesn't. Can you give us a more representative example input file?
I hate excel but I think i can be easily done with "pivot table" isn't it ? https://support.office.com/en-us/article/Calculate-values-in-a-PivotTable-report-11f41417-da80-435c-a5c6-b0185e59da77
Hi Pierre, Thank you so much! I think this is what I need to look for. I will check on it. BTW, is it possible to do so in R? Best, Xiaoyong
@Pierre recommending using excel for something. Well .. that is a first!
No I didn't ! No I didn't ! That wasn't me !
No worries everyone, I made a screenshot.
You are such a meme person!
R dplyr:
group_by->summariseEverything can be done in Excel. Unless you go over 64k lines or whatever the latest Microsoft boss thinks is sufficient. But otherwise, everything.
I agree, Excel is a powerful tool to work with. However, it takes some time to really grab its essence and logic. Since Excel 2007 tables can be "1,048,576 rows by 16,384 columns".
Hmm, I wasn't aware of that. But it seems that Excel is incapable of plotting more that 64k lines in a graph. It doesn't really matter to me how many you can put in a table, if you can't show them in a graph; data that can't be visualized may as well not exist.
"1,048,576 rows by 16,384 columns" is still completely arbitrary. And even if you can paste data in those fields, you can't use it in a graph because Microsoft either lacks skilled programmers or managers. There is no sane reason to restrict people to those limits. If I want to paste a billion lines into Excel, I should be able to do so (if I have sufficient RAM), and it baffles me why Microsoft is incapable of developing tools that can handle the needs of anyone outside of pedestrian users. They are stuck in the "640k is enough for anyone" mindset.
I definitely love the convenience function of converting gene names to dates. Wouldn't know how to work without it!
The only thing you have to do when importing a file is to select the column containing the gene names and set data type to Text.
Yes, that I know. But it's an error which happens easily, silent and irreversible. That shouldn't be possible.
I love Excel for copying and pasting data into a sensible matrix so I can then plot it in something pretty (R - ggplot/plotly) ;) also if you're supervisor is among the less tech savvy, they inevitably want all experiments to produce and excel table they can play with