Error with using WALT

[aganve]

Hello,

I'm in the process of running through the MethPipe pipeline using some practice data provided by your lab. I'm using WALT as recommend for mapping reads. I tried to create an index for hg38 but then got an error saying 'bad index file' so I wasn't able to map my reads. I also used an index available from my lab’s cluster and it gave me the same error. Why am I getting this error? Any ideas on how to fix this?

Verda

[mengzhou]

Hi Verda,

Could you provide the command line you used when encountering this error? This "bad index file" error is reported by walt during mapping. If you have successfully generated the index, it might be that the index file was not found.

[aganve]

So what I tried was:

makedb -c /data/hg38_genome/hg38.fa -o /data/hg38_index/hg38.dbindex

walt -i /data/hg38_index/hg38.dbindex -1 ${INPUT_R1} -2 ${INPUT_R2} -o /DNA_Methylation/2_mapped_reads/${filename}.mr

I've seen that other files were created by the makedb line. Should I be referencing that folder or the specific .dbindex file?

[aganve]

hg38.dbindex_CT00
hg38.dbindex_CT01
hg38.dbindex_GA10
hg38.dbindex_GA11

these were the other files that were created

[mengzhou]

Your command line looks good to me, and I was not able to reproduce this error. In the error message of bad index file, did you notice the file name? For example, if it was bad index file: hg38.dbindex., it might be the index file missing.

[guilhermesena1]

Hi Verda,

If possible could you also clarify whether you cloned the repository or if you are using the release?

Thank you!

[aganve]

I cloned the repository using this command line:
git clone https://github.com/smithlabcode/walt

followed by the instructions you listed in the manual (cd walt, make all, make install). Should I be using the repository or the release? If you see my last post above, I mentioned several .dbindex files were generated when I made my index: hg38.dbindex, hg38.dbindex_CT00, hg38.dbindex_CT01, hg38.dbindex_GA10, hg38.dbindex_GA11

Should I be referencing that folder or the specific .dbindex file when mapping?

[andrewdavidsmith]

@aganve If you could try the same thing with a "release" it might be very helpful. We are having trouble understanding/reproducing the issue.

[aganve]

So I deleted walt and cloned it again. This time I didn't get an error that said 'bad index file.' This solved that problem. Now I have a new issue. I got an emtpy .mr file as well as an empty .mpstats files after mapping. I am using the reads you provided, snippet_1.fq and snippet_2.fq, to map to hg38. No errors came up. Any ideas on how to fix this problem..?

[aganve]

I've used both the repo and release now to try to get my mapped reads but have had 0 success. I'm still getting an empty .mr file. Ideas? Thanks

[guilhermesena1]

Hi Verda,

Can you point us to where you downloaded the hg38 reference? I'm going to try to reproduce the problem. Thank you!

[andrewdavidsmith]

@aganve It would also help if you could send one of us a direct message, and we could provide you with a very small test case file to see if you are able to get success on that file, and if so, whether what you see is identical to what we see when we run it.

[aganve]

I downloaded the hg38 reference from UCSC (https://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/ ). This is what my output error file looks like when mapping to either hg38 or hg19 index:

[MAPPING PAIRED-END READS FROM THE FOLLOWING TWO FILES] /data/hodges_lab/aganve/data/DNA_Methylation/1_FASTQ/snippet_1.fq (AND) /data/hodges_lab/aganve/data/DNA_Methylation/1_FASTQ/snippet_2.fq [OUTPUT MAPPING RESULTS TO /data/hodges_lab/aganve/data/DNA_Methylation/2_mapped_reads/snippet3.mr] The number of reads in paired-end files should be the same.

I am using test data (snippet_1.fq and snippet_2.fq) that was provided with the 2013 version of MethPipe. It looks like rmapbs-pe program was used at that time for mapping. Could there be an issue with the .fq files that doesn't work with walt?

Do you have a test case file you could provide me with? Not sure how to send you a direct message @andrewdavidsmith