Version 2.1 of 'trim_SOLiD_sRNA_cs-fastq.pl' has been released on the 'Tools' page of the website. This version adds the option, at the user's discretion, to keep or remove the 3' "hybrid" color left after initial adapter trimming. In addition, the defaults are now set to output trimmed reads in cs-fastq format, with the 3' hyrbid colors removed. This enables accurate mapping in Colorspace by bowtie 0.12.7 PROVIDED the --col-keepends option is specified (see bowtie 0.12.7 manual). An in-depth discussion of the 3' hybrid color issue in adapter trimming of SOLiD small RNA reads can be found in the README for version 2.1.
On the question of whether it is better to retain trimmed SOLiD small RNA reads in color-space prior to mapping, or to directly translate into DNA first: If there is no reference genome to which your data will be mapped, you are forced to directly translate to make sense of your data. However, if you are mapping the trimmed small RNAs to a reference, my analysis indicates that keeping the reads in color-space will improve the number of mappable reads, especially if you are trying to rescue lower-quality reads by allowing mismatches to the reference. However, it is critical that you understand how your mapping software decodes color-space reads, and the the trimming settings used are compatible.
On the question of whether it is better to retain trimmed SOLiD small RNA reads in color-space prior to mapping, or to directly translate into DNA first: If there is no reference genome to which your data will be mapped, you are forced to directly translate to make sense of your data. However, if you are mapping the trimmed small RNAs to a reference, my analysis indicates that keeping the reads in color-space will improve the number of mappable reads, especially if you are trying to rescue lower-quality reads by allowing mismatches to the reference. However, it is critical that you understand how your mapping software decodes color-space reads, and the the trimming settings used are compatible.