search

mcfrith / last-rna

MIT License
49 stars 6 forks source link

last-train: error: no alignments #13

Open liyonxin opened 7 months ago

liyonxin commented 7 months ago

Hello, I'm comparing environmental microbial samples to the ARG database,but one of samples wrong. When I use the following command : "last-train -P8 -Q0 mydb CH05.fa > reads.train" It reported an error: "lastal: can't calculate E-values." "To proceed without E-values, set a score threshold with option -e." "last-train: error: no alignments" I don't know how to change the value of 'option -e' When I change command: last-train -P48 --revsym --matsym --gapsym -E0.05 -C2 mydb CH05.fa > reads.train It reported : "last-train: error: no alignments" Can you give me some suggestion? Thank you!

mcfrith commented 7 months ago

Hello, I guess it might be the problem mentioned under "Troubleshooting" here: https://gitlab.com/mcfrith/last/-/blob/main/doc/last-train.rst

(The "can't calculate E-values" message is a bit unexpected and confusing, though.)

I hope that helps, Martin

liyonxin commented 7 months ago

hi， thanks for your reply! I have changed command : last-train -P48 -Q0 --sample-number=20000 mydb CH05.fa > reads.train ，but reported same error ： ”lastal: can't calculate E-values“，I have seen the page that you shared before and carefully read it again, but I still don't know how to solve the problem about E-value. I change the last-train'option "-D" value to 1000 ,but it also does't solve. Can you some more detailed solutions?

mcfrith commented 7 months ago

Another possibility is you have an old version: what do you get for lastal --version?

I guess you got a partial reads.train file: could you share it? How big is CH05.fa? What was your lastdb command?

liyonxin commented 7 months ago

hi, I just ran it again, version: 'last from util-linux 2.27.1' lastdb command: lastdb -P48 -uRY4 mydb SARG_20211207_14210_filter.ffn CH05.fa : 5.4G last-train -P48 -Q0 --sample-number=20000 mydb CH05.fa > reads.train reads.train.txt

mcfrith commented 7 months ago

Thanks!

Your "reads.train.txt" file shows that it's lastal version: 1518. It would be best if you could upgrade to at least 1521, which has an improvement to last-train.

However, your "reads.train.txt" shows that last-train found very few alignments (e.g. alignments: 2), which I guess are not real homologies. (In older versions of last-train, if you increase --sample-number, there is a higher risk of getting false homologies. That's what the recent improvement is about.)

It could be that your reads just don't have any homologies to your database.

Another thing is that -uRY4 makes it not very sensitive (but fast and low-memory). If your database isn't huge, you could try omitting -uRY4.

I hope that helps!

liyonxin commented 7 months ago

Thank you very much for your prompt reply！ You are ringt , the data has no homologies with the database. I use conda to update the Last . Since I got your message，I omitted "-uRY4" , "--sample-number" and try again. It took about 15 hours (database 18M) and just finished, the same mistake was reported again. It's really sad. (Just run the end of the program found that the "last_train" version has not changed,but I omitted "--sample-number".) "last-train" found alignments still very few. reads-train.txt