shendurelab / cfdna Goto Github PK
View Code? Open in Web Editor NEWAnalysis of epigenetic signals captured by fragmentation patterns of cell-free DNA
License: MIT License
Analysis of epigenetic signals captured by fragmentation patterns of cell-free DNA
License: MIT License
Hi,
It seems there is line 4 missing in the README file which leads to an absent file that is needed for consecutive processing.
My guess is that it would be something like:
./extractGeneBody.py | bgzip -c > ensemblv75_canonicalTranscriptIDs.protein_coding.lst.gz
Another issue: cfDNA/WPS_overlays/Maurano_et_al_TFclusters/README seems to refer to data/Maurano_et_al_func_var/hg19.taipale.recluster.counts in line 18. How do I get this file?
Thanks,
Jan
I am having some difficulties getting plots.R
to run upon the FFT summary data output by convert_files.py
(everything up to this point worked as expected).
I am using the Protein Atlas file from: http://v13.proteinatlas.org/download/rna.csv.zip. (The link provided in the Human Protein Atlas section of the README is no longer functioning.) It appears that the Protein Atlas file actually used was post-processed, to convert it from comma- to tab-delimited and likely also to rename some columns, based upon the plots.R
script.
The main issue I am currently encountering has to do with the first computation of a correlation, on line 36. The issue is that logndata2
has all NAs, since its assignment on line 34 does not work. This appears to be due to the initial assignment of row names (rownames(proteinAtlas) <- proteinAtlas$GeneID
; line 6) not working, since there is no GeneID
column in my Protein Atlas file.
However, if I change this assignment to instead use the Gene
column, it does not work due to non-unique row names since there are multiple FPKM values per gene ID (one per tissue type or cell line).
If I instead re-write line 34 as follows:
labelledLogndata <- cbind(proteinAtlas[,1], logndata)
logndata2 <- labelledLogndata[labelledLogndata[,1] %in% fdata[,1],]
to not depend upon row names, I end up with a logndata2
dimension of Z x 5, while fdata
has a dimension of Y x 64, where the number of genes are equal, but Z and Y are not. The correlation on line 36 currently uses logndata2
directly, but this results in a non-numeric error. If I instead use logndata2[,3]
, to obtain the FPKM column, I then obtain an incompatible dimension error. Finally on line 39, logndata2[,"NB.4"]
is used, but I do not have any columns with a NB.4
label.
Is there another version of the script which pre-processed the data to resolve this or am I perhaps using an incorrect Protein Atlas file? If you have any suggestions on how to resolve these issues, please let me know.
Hi,
I am currently working on our dataset and especially on the bam file available at http://krishna.gs.washington.edu/download/cfDNA-Nucleosomes/BAMs/ but I can't find CH01.bam on our website.
I am also unable the SRA file that corresponds to this file in the GEO project from NCBI ( GSE71378)
I was only able to get GSE71378_CH01.bb
thanks you
Hi,
The modified samtools provided here could not be use at a CentOS-5.5 cluster. Could you please upload the source code of samtools that we could re-compile on our cluster?
Thanks
Hi,
I followed the instruction on this github to run the scripts to get extracted WPS from FFT. My codes are below:
sample='sample'
count_dir=/wps/body/${sample}/counts/
fft_dir=/wps/body/${sample}/fft/
mkdir -p ${count_dir}
mkdir -p ${fft_dir}
/expression/extractReadStartsFromBAM_Region_WPS.py --minInsert=120 --maxInsert=180 -i /wps/test.tsv -o ${count_dir}/block_%s.tsv.gz sample.cram
cd ${count_dir}
ls block_*.tsv.gz | xargs -n 500 Rscript /expression/fft_path.R ${count_dir} ${fft_dir}
/expression/convert_files.py -a /wps/gene_body.tsv -t /wps/ -r /wps/ -p body -i sample
My plasma samples are ~20x, after I read the results, the numbers are just a bit strange to me.
For example, in the block_*_.tsv.gz, looks like all the WPS is negative, which is unlikely, as it seems to me 5th column = 3rd col - 4 col and followed by smoothing?
1 982004 13 1 -27
1 982003 14 1 -27
1 982002 14 0 -27
1 982001 14 0 -27
1 982000 14 0 -27
1 981999 14 0 -27
1 981998 14 0 -27
1 981997 15 2 -27
1 981996 15 1 -27
1 981995 15 0 -27
1 981994 16 1 -27
After running, FFT, the final WPS is extremly large. Is it normal? Does this WPS need to be normalized in some way or can I just directly use it for downstream analsysis? What are the Cov and Starts here?
Freq Cov Starts WPS
279 10219.5254719195 0.0238497884478577 8598.75313616417
273 5982.58718950388 0.0140443430887118 7952.29769624378
268 7466.81576752813 0.0144896445869867 11721.1384595369
262 8846.48370222107 0.022895061625473 15191.918936982
257 7442.184665902 0.0313783453071702 16884.9712819869
252 5797.68161245892 0.0306839303032613 17429.5203876919
248 14007.9825155327 0.0561427608426257 46347.379064728
243 27006.8319166288 0.0855563232912757 84710.8334345234
239 22558.520706278 0.075053714590048 69932.0207777483
234 13085.7261256437 0.0747390689683382 46922.1465931742
230 10086.7106152956 0.0758667625959573 39388.6401764214
226 5792.06498397896 0.068366103385063 24709.960170311
222 4104.03613727378 0.087971946632555 27390.4829599399
Thanks!
Hi,
I find it very hard to execute your scripts.
Are you working on any packages that allow us to do cf-nucleosome analysis?
Thank you.
Hello,
I've been trying to get the same Canonical Transcript IDs that you uploaded in cfDNA/data/Ensembl_v75/
Though I don't come to the same list when using the above lists, to select the Canonical Transcript IDs. :
Nor do I come to the same list when selecting the longest transcripts.
Would you mind telling me the way you settled for those particular transcripts ?
Did you use particular attributes obtained with BiomaRt to perform your selection ?
Best regards,
MushuW.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.