Very grateful CRISPRseq has added a screening opti...
# crisprseq
i
Very grateful CRISPRseq has added a screening option, fantastic work! 🎊 I'm curious, for MAGeCK rra_contrasts parameter, is there an example of the format of CSV file that is expected? And can this take more than one set of paired files (e.g. if I have 2 treatments each with two replicates and want to compare to two controls, perhaps just as additional lines in the CSV?). Thanks!!
l
Hi Ittai! Thanks for the feedback 🙂 also thanks for the question, I’ll make the documentation clearer!! The csv file needs to have to column : reference and treatment (if you have any better suggestions let me know) :
Copy code
treatment,reference
treatment_replicate_1,control_replicate_1
treatment_replicate_1,control_replicate_2
treatment_replicate_2,control_replicate_2
treatment_replicate_2,control_replicate_1
would this for instance match the comparisons you’d want to do?
I also noticed a tiny bug that i fixed right now so this repo works better (will do a patch asap) https://github.com/LaurenceKuhl/crisprseq/tree/bump
i
Hi Laurence, thanks for your prompt response! That csv structure seems to work. However, I am now encountering a different issue leading to a failure with the MAGECK_TEST process: it complains that the header line of count_table.count.txt only shows 3 dimensions, whereas line 2 shows 6. I looked at that intermediate file output (count_table.count.txt) and it looks like each line should be showing 8 columns. Is it perhaps because some of the column names are identical for different sample replicates (I've attached the head of the file, count_table.sub.txt)? The identical column names represent the different replicates of the same condition (see attached design file, perhaps I need to modify this as well?). Additionally, I believe MAGeCK rra can do pairwise comparison at the condition level taking in multiple samples (replicates) per condition. So the current MAGeCK_TEST call that fails appears as:
Copy code
/mageck test -t treatment50 -c treatment4 -k count_table.count.txt -n treatment50_treatment4
Whereas I would hope for the below, with the attached rra design as well:
Copy code
/mageck test -t treatment50 treatment50 -c treatment4 treatment4 -k count_table.count.txt -n treatment50_treatment4
I presume this would present an issue with respect to specifying the correct columns but am curious how I would go about performing grouped RRA with the correct groupings specified in that case, if possible. Thanks!!
👀 1
l
Could you by any chance try with the following files :
rra_design.csv,crispr_design_template.csv
I found this in the mageck doc :
Copy code
For example in the test command, "-t sample1_bio_replicate1,sample1_bio_replicate2 -c sample2_bio_replicate1,sample2_bio_replicate2" compares 2 samples (with 2 biological replicates in each sample).
i
Thanks for your quick reply and these additional files! I tried running with these and note that I encountered an issue still with the csv splitting of rra design, from the task processes it appeared that only the first csv was being split (i.e. quotes were not successfully delimiting the paired csvs from one another). I changed the "sep" used in the splitCsv call for parsing rra_contrasts before the MAGECK_TEST process to use ";" and changed the rra design file correspondingly (i.e. semicolon-delimited header & rows with semicolon-delimited csv lists of control & treatment samples) and found that this worked. Thanks again!
l
Hi Ittai, thanks a lot for keeping me posted, i’ll modify the code and the documentation for the next release 🙂
🙌 1
the next release should also have a few graphs checking sample replicates similarity! 🎉
🙌 1
a
Hi, I have tried to run the crispr-seq screening with three treatment groups and 2 controls. The mle folder outputs few file and looking at the count summary, I only get the output with three treated samples. I am not sure why my negative control didn't get mapped? What do you suspect? Thank you